Meeting Title: DE-AE-AI Standup Date: 2025-11-26 Meeting participants: Samuel Roberts, Uttam Kumaran, Mustafa Raja, Rico Rejoso, Casie Aviles, Demilade Agboola, Gabriel Lam, Awaish Kumar
WEBVTT
1 00:00:18.420 ⇒ 00:00:19.720 Uttam Kumaran: Hey, good morning.
2 00:00:25.650 ⇒ 00:00:26.760 Samuel Roberts: Can you hear me now?
3 00:00:26.960 ⇒ 00:00:27.670 Uttam Kumaran: Yes.
4 00:00:28.270 ⇒ 00:00:31.490 Samuel Roberts: Oh, boy, but I can’t hear you, excellent. It’s never, never as good.
5 00:00:32.630 ⇒ 00:00:35.330 Samuel Roberts: My computer hates when I change setups.
6 00:00:37.630 ⇒ 00:00:38.550 Mustafa Raja: Hmm…
7 00:00:39.670 ⇒ 00:00:40.550 Uttam Kumaran: Hey!
8 00:00:41.840 ⇒ 00:00:43.850 Mustafa Raja: I met with Ashwini today.
9 00:00:44.470 ⇒ 00:00:45.700 Uttam Kumaran: Oh, nice, how’d it go?
10 00:00:46.430 ⇒ 00:00:54.039 Mustafa Raja: Yeah, super nice. Yeah, we might… we might meet, again. He had… he needed the Snowflake access.
11 00:00:54.220 ⇒ 00:00:55.580 Mustafa Raja: the CT and all.
12 00:00:55.920 ⇒ 00:01:01.370 Mustafa Raja: But I had some questions about dbt and, yeah, he cleared some up.
13 00:01:01.830 ⇒ 00:01:05.420 Mustafa Raja: We might meet on Friday or Thursday, again.
14 00:01:06.360 ⇒ 00:01:06.900 Uttam Kumaran: Okay.
15 00:01:09.260 ⇒ 00:01:13.140 Mustafa Raja: Do we have stand-ups for Friday and Thursday? It’s Thanksgiving, right?
16 00:01:13.630 ⇒ 00:01:17.060 Uttam Kumaran: I’m gonna leave it, but I won’t be attending.
17 00:01:18.150 ⇒ 00:01:20.450 Mustafa Raja: Okay, I might work in my time zone, then.
18 00:01:20.630 ⇒ 00:01:21.180 Uttam Kumaran: Okay.
19 00:01:31.050 ⇒ 00:01:32.260 Samuel Roberts: I think I’m good enough.
20 00:01:33.090 ⇒ 00:01:34.239 Samuel Roberts: How’s everyone?
21 00:01:34.520 ⇒ 00:01:35.500 Uttam Kumaran: Hey, good.
22 00:01:36.860 ⇒ 00:01:41.119 Samuel Roberts: different headphones today, and it doesn’t like it. Although it liked it yesterday, but I don’t know.
23 00:01:55.540 ⇒ 00:01:56.770 Uttam Kumaran: For a couple others.
24 00:02:07.430 ⇒ 00:02:11.780 Mustafa Raja: So, what’s the reasoning behind avoiding the staging layer in dbt?
25 00:02:16.010 ⇒ 00:02:18.300 Uttam Kumaran: What’s the reason behind… say it again?
26 00:02:19.330 ⇒ 00:02:26.730 Mustafa Raja: What’s the reasoning behind avoiding the staging layer in the BT we got from raw to… Intermediate, right?
27 00:02:27.740 ⇒ 00:02:33.879 Uttam Kumaran: Oh, so there shouldn’t… so staging is going to be, like, an environment, not a layer.
28 00:02:34.470 ⇒ 00:02:35.840 Uttam Kumaran: Typically.
29 00:02:36.240 ⇒ 00:02:40.450 Uttam Kumaran: So, I mean, it depends, like, I think Ashwini has a little bit of a different…
30 00:02:40.640 ⇒ 00:02:46.930 Uttam Kumaran: naming convention, but typically you have raw, intermediate, marts?
31 00:02:47.200 ⇒ 00:02:51.409 Uttam Kumaran: And then you can have a dev, staging, and production environment.
32 00:02:52.000 ⇒ 00:02:52.770 Mustafa Raja: Huh.
33 00:02:53.990 ⇒ 00:02:56.580 Samuel Roberts: Would those environments normally have, like, test data or something? Is that…
34 00:02:56.920 ⇒ 00:03:06.909 Uttam Kumaran: Well, it’s more about, like, the stage in which your code is. So, like, while you’re testing and development, like, you write a SQL query, you want to run it.
35 00:03:07.110 ⇒ 00:03:23.670 Uttam Kumaran: you want it to materialize in a dev environment. When you push a PR, you want that to materialize in the staging environment. So, for example, other people can go into staging and access. So you… so typically, you have one dev schema or environment per
36 00:03:23.970 ⇒ 00:03:29.050 Uttam Kumaran: team member. And that way, it allows you to, like.
37 00:03:29.340 ⇒ 00:03:31.899 Samuel Roberts: To just develop and kind of see the flow.
38 00:03:31.900 ⇒ 00:03:42.120 Uttam Kumaran: And then, once in staging, other people can check. Sometimes you may have staging dashboards or other tests that run, and then you have a production. Additionally, you may even have a pre-production.
39 00:03:42.250 ⇒ 00:03:46.760 Uttam Kumaran: Which is like, okay, changes are made, and then you do a deploy, you know?
40 00:03:47.140 ⇒ 00:03:48.549 Samuel Roberts: Okay, okay, yeah.
41 00:03:48.860 ⇒ 00:03:54.410 Samuel Roberts: I’m used to, like, staging and dev, having, like, test data and dummy data for, like, developing the, you know.
42 00:03:54.650 ⇒ 00:03:56.170 Samuel Roberts: SaaS app kind of thing.
43 00:03:56.650 ⇒ 00:04:00.929 Samuel Roberts: But I guess when it’s data, you want the data, you don’t just want test data.
44 00:04:01.110 ⇒ 00:04:01.810 Uttam Kumaran: Yeah.
45 00:04:04.460 ⇒ 00:04:07.660 Uttam Kumaran: Let me think… Cool.
46 00:04:15.130 ⇒ 00:04:18.939 Uttam Kumaran: Maybe we can start talking, Hedra?
47 00:04:19.350 ⇒ 00:04:21.339 Uttam Kumaran: Demolade, maybe we can start there.
48 00:04:23.360 ⇒ 00:04:27.260 Demilade Agboola: Yes, I sent my PR in, I have done the…
49 00:04:28.080 ⇒ 00:04:36.500 Demilade Agboola: Modification, so we can see the cancel dates, as long as they’re not currently active subscription.
50 00:04:36.910 ⇒ 00:04:45.860 Demilade Agboola: And also we’ve added, like, the survey information as well, so the latest survey they filled, and the other information attached to the surveys.
51 00:04:46.250 ⇒ 00:04:49.879 Demilade Agboola: Yeah, it’s all in Dim users right now.
52 00:04:50.890 ⇒ 00:04:51.440 Uttam Kumaran: Okay.
53 00:04:52.540 ⇒ 00:05:01.950 Demilade Agboola: And also, yeah, I also flagged the fact that there were duplicates in the… because the test was fair, and I was curious as to why. So I kind of dug into that. So it turns out Stripe is sending
54 00:05:02.120 ⇒ 00:05:03.760 Demilade Agboola: Stripe has duplicates.
55 00:05:04.090 ⇒ 00:05:21.029 Demilade Agboola: And so when we joined the in-house data to the Stripe data, it causes 14 records to duplicate into about 28 or 30 records. And so once I sent that to their team, they said it seems like it’s a bad immigration, like, because they immigrated to Stripe, so…
56 00:05:21.410 ⇒ 00:05:22.300 Uttam Kumaran: Whoa.
57 00:05:22.300 ⇒ 00:05:33.440 Demilade Agboola: data, so that’s kind of why. But they’re not concerned, because, you know, they have way… like, 14 is a very small number relative to the size of customers they have, so they’re cool with that.
58 00:05:33.850 ⇒ 00:05:34.440 Uttam Kumaran: Okay.
59 00:05:38.160 ⇒ 00:05:43.629 Uttam Kumaran: Okay, great. Did you happen to let… does… is Hedra aware of, like, the Stripe thing?
60 00:05:44.200 ⇒ 00:05:55.940 Demilade Agboola: Yeah, yeah, I flagged it on there. That’s why they looked into it and said it was this problem, migration on data. So, it’s not on us to fix, it’s on them to fix. So, once that’s test should stop failing.
61 00:05:56.420 ⇒ 00:06:02.370 Demilade Agboola: But I was thinking about trying to test so that I can flag those 14 records, I would exclude them.
62 00:06:02.800 ⇒ 00:06:09.359 Demilade Agboola: That way, if we get new duplicates in the future, that’s when it should… that’s when the triggering should happen.
63 00:06:10.170 ⇒ 00:06:13.020 Uttam Kumaran: Yeah, up to you. Yeah.
64 00:06:15.300 ⇒ 00:06:22.419 Uttam Kumaran: You can do… you can… you can honestly do one test just for those 14, and then one test for not… anything not in those 14.
65 00:06:23.120 ⇒ 00:06:23.740 Uttam Kumaran: Yeah.
66 00:06:24.000 ⇒ 00:06:27.460 Uttam Kumaran: If that’s… that way, you don’t… you don’t have to introduce logic.
67 00:06:28.660 ⇒ 00:06:37.969 Demilade Agboola: Yeah, yeah, I’ll probably just do, like, yeah, excluding those rows, I don’t know, yeah, those records, and that should be fine. Test should stop failing.
68 00:06:38.490 ⇒ 00:06:43.910 Demilade Agboola: But yeah, everything’s good. I don’t know if there’s any other thing we need to push out for them.
69 00:06:45.430 ⇒ 00:06:47.949 Uttam Kumaran: I think that’s it. I think, just…
70 00:06:48.170 ⇒ 00:06:52.289 Uttam Kumaran: I know Awash sent some stuff to Hydra today, so…
71 00:06:52.840 ⇒ 00:06:56.130 Uttam Kumaran: I will send a ping to Sandra as well, saying.
72 00:06:56.790 ⇒ 00:07:04.549 Uttam Kumaran: let me know what else, where we can support. The dbt stuff, maybe, Demi, just glance at that so you can…
73 00:07:05.280 ⇒ 00:07:09.419 Uttam Kumaran: get a sense of it, but I’ve set everything up kind of… we set everything up kind of normally.
74 00:07:09.630 ⇒ 00:07:10.600 Uttam Kumaran: Yep.
75 00:07:11.890 ⇒ 00:07:12.770 Uttam Kumaran: Yeah.
76 00:07:12.770 ⇒ 00:07:17.139 Demilade Agboola: I noticed some job wrong with failing, but I’ll look into that today.
77 00:07:17.330 ⇒ 00:07:18.830 Uttam Kumaran: I think it’s the tests.
78 00:07:20.150 ⇒ 00:07:25.160 Demilade Agboola: Okay, fair enough. I’ll look into those as well, just to kind of get an idea of what’s happened.
79 00:07:25.960 ⇒ 00:07:26.560 Uttam Kumaran: Okay.
80 00:07:32.430 ⇒ 00:07:37.629 Uttam Kumaran: Okay, great. Maybe Mustafa, any progress on README yesterday?
81 00:07:39.470 ⇒ 00:07:40.160 Uttam Kumaran: Or anything.
82 00:07:40.160 ⇒ 00:07:46.279 Mustafa Raja: Yes, on, on really, I met with a wish, to figure out…
83 00:07:46.400 ⇒ 00:07:57.580 Mustafa Raja: the end users, that they, they are labeling. So there wasn’t a good identifier. We got a response, from one of the engineers, I believe, today.
84 00:07:57.700 ⇒ 00:08:14.310 Mustafa Raja: And, yeah, we, we do now know, where, where the permissions, to those, projects live, so I might be able to, bring the sign-up usage number down a little.
85 00:08:14.430 ⇒ 00:08:24.150 Mustafa Raja: But they don’t have a good identifier if they are only signing up to suggest edits for public projects rather than the private ones, so…
86 00:08:24.250 ⇒ 00:08:32.220 Mustafa Raja: Yeah. I’ll do a new aggregation, based on their feedback.
87 00:08:32.360 ⇒ 00:08:39.529 Mustafa Raja: Other than that, they’re saying that the, amplitude event for user sign-up success is firing.
88 00:08:39.700 ⇒ 00:08:41.900 Mustafa Raja: As expected.
89 00:08:42.240 ⇒ 00:08:46.810 Mustafa Raja: I didn’t see that, when I created two new accounts to test that.
90 00:08:47.050 ⇒ 00:08:51.170 Mustafa Raja: But I guess if it’s working for them, then that’s okay.
91 00:08:51.610 ⇒ 00:08:52.200 Uttam Kumaran: Okay.
92 00:08:55.110 ⇒ 00:09:00.679 Uttam Kumaran: Okay, yeah, I’ll let you… I think you got it, so… nothing from me, yeah.
93 00:09:03.250 ⇒ 00:09:06.069 Mustafa Raja: Yeah, I’m in sync with, Robert on all these updates.
94 00:09:06.550 ⇒ 00:09:07.100 Uttam Kumaran: Okay.
95 00:09:11.740 ⇒ 00:09:19.580 Uttam Kumaran: Okay, great. Maybe ABC stuff? Anything from yesterday?
96 00:09:22.810 ⇒ 00:09:32.670 Casie Aviles: From yesterday, there’s not… yeah, there’s not much with ABC, I think just the… The plan,
97 00:09:33.480 ⇒ 00:09:38.430 Casie Aviles: Yeah, I think if… if that’s… If there’s anything else missing…
98 00:09:38.860 ⇒ 00:09:46.849 Casie Aviles: Just… yeah, I think that’s all I need to know, if there’s still anything else that I need to work on that, but…
99 00:09:47.200 ⇒ 00:09:47.590 Uttam Kumaran: Okay.
100 00:09:47.590 ⇒ 00:09:56.169 Casie Aviles: Yeah, it’s pretty much there. So, yeah, just waiting on further review, if there’s anything else, any comments.
101 00:09:57.000 ⇒ 00:09:59.609 Uttam Kumaran: Yeah, so let me put in my calendar to review.
102 00:10:00.270 ⇒ 00:10:07.659 Samuel Roberts: Yeah, I spent a little time looking at, like, GCP stuff, and I think the way we have it set up with the cloud run should work with Maestra, from what I can understand, so…
103 00:10:07.940 ⇒ 00:10:08.480 Uttam Kumaran: Okay.
104 00:10:10.960 ⇒ 00:10:19.130 Samuel Roberts: it doesn’t have, like, an automatic… like, some… they have other serverless deployments kind of set up in their documentation, but I think we can do it based on some blog posts I saw.
105 00:10:19.420 ⇒ 00:10:20.330 Samuel Roberts: Okay.
106 00:10:22.880 ⇒ 00:10:28.890 Samuel Roberts: And it… the way they want us to do it is, like, containers, so that should make migrating to their services even easier.
107 00:10:29.170 ⇒ 00:10:29.770 Uttam Kumaran: Okay.
108 00:10:30.300 ⇒ 00:10:32.139 Samuel Roberts: I’m pretty… I think that’ll be good.
109 00:10:35.590 ⇒ 00:10:36.200 Uttam Kumaran: Okay.
110 00:10:36.500 ⇒ 00:10:37.210 Uttam Kumaran: Great.
111 00:10:38.230 ⇒ 00:10:40.690 Uttam Kumaran: What else?
112 00:10:43.240 ⇒ 00:10:46.789 Uttam Kumaran: I guess, how’s it going, Casey, on Eden’s stuff?
113 00:10:46.940 ⇒ 00:10:51.169 Uttam Kumaran: Anything on, like, the data side that, need help with?
114 00:10:52.240 ⇒ 00:11:00.559 Casie Aviles: Yeah, so for Eden, yeah, I tried doing one of the analysis work, so I was just…
115 00:11:00.780 ⇒ 00:11:06.460 Casie Aviles: I met with, Henry yesterday, he briefed me on, like, what we needed to do, so…
116 00:11:06.840 ⇒ 00:11:10.700 Casie Aviles: Basically, I’m just handling, like, getting…
117 00:11:11.420 ⇒ 00:11:17.900 Casie Aviles: Refund… catalyst refunds, and, like, looking at…
118 00:11:18.320 ⇒ 00:11:25.270 Casie Aviles: Basically, I was trying to create a table that looks at…
119 00:11:25.890 ⇒ 00:11:30.099 Casie Aviles: Records that are, you know, first touch and last touch.
120 00:11:30.210 ⇒ 00:11:32.090 Casie Aviles: Catalyst, and then…
121 00:11:33.170 ⇒ 00:11:38.290 Casie Aviles: So that’s what I was trying to do, but there was… there’s, like, a very low amount of…
122 00:11:39.370 ⇒ 00:11:49.579 Casie Aviles: that… there’s a very low count of that, so I wanted to do a sanity check first, and I asked him to, like, take a look if I didn’t… if ever I missed anything.
123 00:11:50.440 ⇒ 00:11:54.800 Casie Aviles: But yeah, the end goal is to be able to compare, like, the…
124 00:11:55.340 ⇒ 00:11:58.430 Casie Aviles: The rate of, you know, the refunds.
125 00:11:58.590 ⇒ 00:12:00.570 Casie Aviles: With non-catalysts.
126 00:12:00.810 ⇒ 00:12:05.409 Casie Aviles: But, yeah, that’s… that’s all I have from yesterday.
127 00:12:05.880 ⇒ 00:12:06.520 Uttam Kumaran: Okay.
128 00:12:09.010 ⇒ 00:12:12.390 Casie Aviles: Yeah, I’ll be continuing work on that today.
129 00:12:12.390 ⇒ 00:12:12.940 Uttam Kumaran: Okay.
130 00:12:16.020 ⇒ 00:12:16.750 Uttam Kumaran: Great.
131 00:12:16.970 ⇒ 00:12:19.999 Uttam Kumaran: So I’ll let you work with him on that. Cool.
132 00:12:21.150 ⇒ 00:12:23.709 Uttam Kumaran: I guess. Otherwise,
133 00:12:24.330 ⇒ 00:12:32.299 Uttam Kumaran: Don’t have any… I don’t think there’s anything else on any other clients. I’ll chat with, Sweeney directly about DTA.
134 00:12:32.860 ⇒ 00:12:37.950 Uttam Kumaran: I guess anything else, Demi, on, like, the Eden side for…
135 00:12:38.350 ⇒ 00:12:41.789 Uttam Kumaran: data. Oisha’s in an interview right now, so…
136 00:12:41.790 ⇒ 00:12:46.990 Demilade Agboola: Gotcha. So for Eden… We’re trying to get the offline status in.
137 00:12:47.170 ⇒ 00:12:56.159 Demilade Agboola: But we don’t have the API credentials for that. What the Edensim sent over doesn’t work.
138 00:12:56.420 ⇒ 00:13:02.950 Demilade Agboola: And so we’ve connected to the, like, Polytomic and ourselves have connected to the Applian’s team.
139 00:13:03.290 ⇒ 00:13:07.419 Demilade Agboola: We’ve given them what the issues are, and they said they’ll connect
140 00:13:07.520 ⇒ 00:13:12.309 Demilade Agboola: Internally, and let us know. So, we’re kind of waiting for that to get to go ahead to…
141 00:13:12.720 ⇒ 00:13:19.340 Demilade Agboola: build out the API, and once that’s done, we can start to use the upfronts data for the error.
142 00:13:19.510 ⇒ 00:13:21.400 Demilade Agboola: Marketing Dashboard Reports.
143 00:13:22.070 ⇒ 00:13:22.640 Uttam Kumaran: Okay.
144 00:13:29.970 ⇒ 00:13:31.370 Uttam Kumaran: Okay, great.
145 00:13:35.670 ⇒ 00:13:46.590 Uttam Kumaran: Okay, anything else? So, I guess, Mustafa, you just met with Ashwini about the dbt. Did you guys edit that doc at all? I have time, I can review it today, this morning.
146 00:13:46.790 ⇒ 00:13:53.080 Mustafa Raja: No, we didn’t, Ashwini needed access on, Snowflake.
147 00:13:53.770 ⇒ 00:14:00.919 Mustafa Raja: So, to work on it, or, initiate dbt,
148 00:14:01.100 ⇒ 00:14:03.629 Mustafa Raja: He wanted that access first, so…
149 00:14:04.110 ⇒ 00:14:06.909 Uttam Kumaran: To, like, to run through and, like, test things?
150 00:14:07.100 ⇒ 00:14:07.880 Mustafa Raja: Yeah.
151 00:14:08.140 ⇒ 00:14:09.300 Uttam Kumaran: Okay, okay, okay.
152 00:14:10.310 ⇒ 00:14:17.799 Mustafa Raja: Yeah, also, I don’t know if that was, if that was for our internal Snowflake, or if CTA has their own Snowflake.
153 00:14:18.320 ⇒ 00:14:26.959 Uttam Kumaran: Yeah, he wants… well, CTA has one, but that one I have to get access… I have to get him access to, but for the purposes of the dock, you can use our internal one to test
154 00:14:28.200 ⇒ 00:14:32.869 Mustafa Raja: Okay. Yeah, I asked him to, ask Rico to, add him in internally.
155 00:14:32.870 ⇒ 00:14:37.269 Uttam Kumaran: So this is actually probably a good time. I’ll show… I can show you how to do it.
156 00:14:37.750 ⇒ 00:14:39.750 Uttam Kumaran: How to go out of people there. Yeah.
157 00:14:41.030 ⇒ 00:14:45.289 Uttam Kumaran: Okay, cool. I guess let’s, maybe we could…
158 00:14:45.500 ⇒ 00:14:48.320 Uttam Kumaran: Especially talking about internal AI, Gabe.
159 00:14:51.070 ⇒ 00:14:57.300 Gabriel Lam: Yeah, we were able to deploy something yesterday. I think there’s still a couple things for us to…
160 00:14:57.590 ⇒ 00:15:01.499 Gabriel Lam: hammer out and refine, just because…
161 00:15:01.890 ⇒ 00:15:05.840 Gabriel Lam: We’re still testing and making sure some of the pumps are really airtight.
162 00:15:06.090 ⇒ 00:15:13.649 Gabriel Lam: I know the team is, at least the last couple days, we’re focusing on client stuff and getting that out the door.
163 00:15:13.860 ⇒ 00:15:19.870 Gabriel Lam: But we were able to… have the… like…
164 00:15:20.120 ⇒ 00:15:22.900 Gabriel Lam: Pretty robust summaries and tickets coming out.
165 00:15:23.500 ⇒ 00:15:25.319 Gabriel Lam: We were also able to move
166 00:15:26.060 ⇒ 00:15:35.210 Gabriel Lam: some things into length views for additional testing, but yeah, I think generally I’m interested in, you know, whether the team has the capacity today to really
167 00:15:37.110 ⇒ 00:15:39.209 Gabriel Lam: You know, go through the final stretch.
168 00:15:39.520 ⇒ 00:15:40.090 Uttam Kumaran: Okay.
169 00:15:42.840 ⇒ 00:15:43.560 Uttam Kumaran: Cool.
170 00:15:44.810 ⇒ 00:15:51.499 Uttam Kumaran: Okay, great. Yeah, I think, I feel like, is the length views, like, sort of process working? Like, I know we…
171 00:15:51.650 ⇒ 00:15:54.729 Uttam Kumaran: We arrived at, like, okay, we’re gonna… for… for…
172 00:15:54.830 ⇒ 00:15:57.450 Uttam Kumaran: Developing, we’re just gonna do stuff in…
173 00:15:58.000 ⇒ 00:16:04.560 Uttam Kumaran: the repo until we feel comfortable, and then move it to LangFuse? I guess, like, how do you feel it’s working right now?
174 00:16:07.420 ⇒ 00:16:09.859 Gabriel Lam: I am of the position, actually, that
175 00:16:10.100 ⇒ 00:16:13.939 Gabriel Lam: Langfuse offers a better way to test things for
176 00:16:14.270 ⇒ 00:16:18.380 Gabriel Lam: People who are maybe less technical and don’t have to go through the whole repo process.
177 00:16:18.490 ⇒ 00:16:26.070 Gabriel Lam: I don’t know what the cost of that is in either memory, or time or money, and maybe the team has a better idea there.
178 00:16:26.340 ⇒ 00:16:33.300 Gabriel Lam: nor… and I think maybe there’s a conversation about…
179 00:16:33.510 ⇒ 00:16:38.239 Gabriel Lam: Yeah, whether changes in length use, actually, or the velocity in which changes in length use.
180 00:16:39.090 ⇒ 00:16:43.410 Gabriel Lam: Lead to updates in our platform or our deployments.
181 00:16:45.180 ⇒ 00:16:48.519 Gabriel Lam: if Sam or Mustafa or Casey are able to…
182 00:16:50.150 ⇒ 00:16:52.879 Samuel Roberts: I think they should be, like, pretty immediate.
183 00:16:52.880 ⇒ 00:16:53.480 Gabriel Lam: Okay.
184 00:16:53.720 ⇒ 00:16:56.469 Samuel Roberts: I don’t think it’s caching anything, is it? .
185 00:16:57.090 ⇒ 00:16:58.019 Mustafa Raja: Yeah, no, it’s not.
186 00:16:58.020 ⇒ 00:16:58.650 Casie Aviles: Yeah.
187 00:16:58.910 ⇒ 00:17:08.679 Samuel Roberts: Yeah, so when changes are made on LangFuse, they should be the new… like, that prompt… I think you… yeah, you can version on LangFuse and stuff, but I think whatever the current, like, set one is, is what’s gonna get called.
188 00:17:08.890 ⇒ 00:17:10.599 Samuel Roberts: By the… by the 4th.
189 00:17:12.790 ⇒ 00:17:13.490 Gabriel Lam: Great.
190 00:17:15.280 ⇒ 00:17:21.249 Samuel Roberts: And yeah, I think the current process, at least in my mind, works pretty well, where, yeah, developing locally.
191 00:17:21.599 ⇒ 00:17:40.759 Samuel Roberts: developing in GitHub, in code, in Markdown files, or whatever is fine. And then, I don’t think getting things in Lankuse is a huge lift once we’re done and things are stable, so I think it’s pretty… pretty good moving forward. And then, yeah, I don’t know if anyone’s been using it to iterate or not, but, I don’t know how that is on Lankuse, because I haven’t touched that as much.
192 00:17:40.990 ⇒ 00:17:47.569 Samuel Roberts: But then there’s also more features I think we could be making use of, probably for testing and, evaluations and things.
193 00:17:49.540 ⇒ 00:17:50.190 Uttam Kumaran: Okay.
194 00:17:51.980 ⇒ 00:17:54.199 Gabriel Lam: I’ve been using it to iterate, I think it’s just…
195 00:17:54.830 ⇒ 00:18:02.559 Gabriel Lam: In my perspective, it’s a lot faster to do it there, once we’re able to get to that stage, instead of having to go through a few apps and then…
196 00:18:02.700 ⇒ 00:18:05.069 Gabriel Lam: You know… PRs and reviews and all.
197 00:18:05.070 ⇒ 00:18:12.399 Uttam Kumaran: Oh, cause, yeah, because I think, Sam, it’s part of it is, like, the review process requires, like.
198 00:18:12.940 ⇒ 00:18:18.160 Uttam Kumaran: The UI may require changes based on the structure of the… the AI output.
199 00:18:18.880 ⇒ 00:18:23.100 Uttam Kumaran: And so, maybe it’s, like… Yeah.
200 00:18:24.570 ⇒ 00:18:40.490 Uttam Kumaran: kind of see what I’m saying, like, it’s… if… we may be… like, the… probably the iteration process Gabe’s talking about is, like, oh, the output in the review app isn’t great, okay, now make another change, whereas Gabe can just make the prompt changes and link views until it lines up.
201 00:18:40.700 ⇒ 00:18:45.829 Samuel Roberts: But even in, even in, like, dev, kind of… Situation, where we’re still refining
202 00:18:46.510 ⇒ 00:18:53.470 Samuel Roberts: the process… well, I guess, yeah, I mean, once the code is locked pretty well, then that should be fine.
203 00:18:54.100 ⇒ 00:19:00.790 Uttam Kumaran: I guess, is that… is that right, Gabe? Like, the prompt… the output changes so much that it may require UI changes?
204 00:19:01.110 ⇒ 00:19:03.610 Uttam Kumaran: And so… Yeah.
205 00:19:04.000 ⇒ 00:19:05.890 Gabriel Lam: I think… Hmm.
206 00:19:09.930 ⇒ 00:19:12.720 Gabriel Lam: I don’t want to say it depends, but it kind of depends.
207 00:19:14.290 ⇒ 00:19:16.179 Samuel Roberts: No, I think you’re right.
208 00:19:16.180 ⇒ 00:19:18.580 Gabriel Lam: I think, generally, if they’re more about
209 00:19:19.730 ⇒ 00:19:23.009 Gabriel Lam: Refining and noticing, like, hey, you know, the output
210 00:19:23.030 ⇒ 00:19:40.539 Gabriel Lam: needs to hit a certain structure, or we’re missing certain pieces, like, key pieces of information in the output, then I think that’s fine. I think if we’re going in and, like, you know, throwing a hammer into it and totally changing certain things, then obviously that results in UI changes that are pretty drastic.
211 00:19:40.800 ⇒ 00:19:46.480 Gabriel Lam: But, like, the summary per video, for example, I think that is a lower…
212 00:19:46.860 ⇒ 00:19:47.430 Samuel Roberts: Yeah.
213 00:19:49.520 ⇒ 00:19:59.469 Gabriel Lam: a smaller lift in terms of development, or, like, the likelihood of changes to the UI there are lower, and I think in cases like that.
214 00:19:59.650 ⇒ 00:20:05.290 Gabriel Lam: I find it more productive if we’re able to get people’s, input.
215 00:20:05.910 ⇒ 00:20:11.329 Gabriel Lam: Maybe through Langviews, as opposed to going through a whole, review and deploy and PR process.
216 00:20:12.520 ⇒ 00:20:21.210 Samuel Roberts: Yeah, I think… I think you’re absolutely right. If it’s less structured data and just text that’s coming out, then it’s not… then you don’t have to worry about how it’s gonna change,
217 00:20:21.460 ⇒ 00:20:29.970 Samuel Roberts: So we might just need to keep that in mind when we’re… if there’s things that are creating structured text in the future, or currently, tweaking those prompts is… is a little…
218 00:20:30.660 ⇒ 00:20:35.989 Samuel Roberts: more fragile or finicky to just doing LangFuse automatically.
219 00:20:37.900 ⇒ 00:20:47.539 Samuel Roberts: So maybe, I mean, I guess what I’m saying is, like, yeah, things in line queues should definitely be ones that are just text outputs, like summaries, or chat responses, or things like that.
220 00:20:47.740 ⇒ 00:20:51.949 Samuel Roberts: Maybe we need to then consider if things are a little more structured.
221 00:20:52.290 ⇒ 00:20:57.379 Samuel Roberts: What we do, whether it’s in length views or not, because changing those could break things pretty significantly.
222 00:20:57.840 ⇒ 00:20:58.400 Gabriel Lam: Yeah.
223 00:20:58.480 ⇒ 00:20:59.400 Samuel Roberts: Okay.
224 00:21:04.240 ⇒ 00:21:13.990 Uttam Kumaran: I mean, yeah, for me, like, there… you know this, we have the sidebar agents? Like, one of the things that I wanted to try to get to, and I have a little bit of time, is, like, moving some of my ChatGPT prompts.
225 00:21:14.080 ⇒ 00:21:24.469 Uttam Kumaran: Into there, and then I can start to tweak them, but… yeah, I guess, like, for the PR process, like, should I just go ahead and try to move those to LangFuse?
226 00:21:25.350 ⇒ 00:21:26.010 Uttam Kumaran: Nope.
227 00:21:26.010 ⇒ 00:21:30.229 Samuel Roberts: It’s an interesting case, because those are stored in Supabase right now, so you can go and add agents.
228 00:21:30.400 ⇒ 00:21:35.009 Samuel Roberts: and prompts, right now, without any PR or anything. It’s all just data.
229 00:21:35.010 ⇒ 00:21:39.250 Uttam Kumaran: But the thing about… the thing about LangFuse is just, like, that’s the iteration. Like, I want to be able to.
230 00:21:39.250 ⇒ 00:21:41.230 Samuel Roberts: No, I agree, I’m saying, I think, I think we…
231 00:21:41.230 ⇒ 00:21:41.930 Uttam Kumaran: I hear you, I hear you.
232 00:21:42.140 ⇒ 00:21:43.130 Samuel Roberts: the…
233 00:21:43.400 ⇒ 00:22:00.310 Samuel Roberts: basically right now, instead of it being… living in code like some of the prompts did for certain parts of the forge, those agents live in Superbase, so moving those over to Langviews is pretty easy at that point. But we’d have to just change how we’re fetching things. Instead of being fetching from Superbase, it would be fetching from Langviews.
234 00:22:00.310 ⇒ 00:22:10.229 Samuel Roberts: And then, I assume, and Casey Mustafa, let me know, that there’s a way to add prompts programmatically with the API, so that we could still have the, like, add agent function?
235 00:22:10.530 ⇒ 00:22:14.279 Samuel Roberts: on the forge, and not just rely purely on LangFuse’s UI.
236 00:22:14.380 ⇒ 00:22:16.480 Samuel Roberts: But I don’t know that.
237 00:22:17.020 ⇒ 00:22:17.470 Mustafa Raja: Yes.
238 00:22:17.470 ⇒ 00:22:19.220 Uttam Kumaran: Basically, I think, like, a lot of folks…
239 00:22:19.220 ⇒ 00:22:19.900 Mustafa Raja: in the…
240 00:22:19.900 ⇒ 00:22:23.449 Uttam Kumaran: A lot of the users, I want to turn into, like, more prompt…
241 00:22:23.450 ⇒ 00:22:23.970 Samuel Roberts: Right.
242 00:22:23.970 ⇒ 00:22:25.430 Uttam Kumaran: Adjusters, you know?
243 00:22:25.430 ⇒ 00:22:31.990 Samuel Roberts: Yeah, the question, I think, becomes, do you want to have those prompts editable on the… on the Forge at all, or just in link views?
244 00:22:34.770 ⇒ 00:22:35.740 Uttam Kumaran: Hmm…
245 00:22:36.020 ⇒ 00:22:44.130 Samuel Roberts: Because right now, you can change those prompts and test on the forge. The flow is not necessarily ideal, because you have to go to, like, the agents, but we could make that, you know.
246 00:22:44.340 ⇒ 00:22:55.450 Samuel Roberts: a little more… especially for something that’s just a chat thing, and it’s just text. We lose the kind of versioning, and potentially, like, the eval stuff that we might get from LangFuse in the future, but…
247 00:22:58.100 ⇒ 00:23:01.549 Samuel Roberts: You know, right now, if you want to add an agent or change an agent, that’s all doable.
248 00:23:01.970 ⇒ 00:23:03.089 Samuel Roberts: In the forge.
249 00:23:04.780 ⇒ 00:23:05.500 Uttam Kumaran: Hmm, okay.
250 00:23:05.500 ⇒ 00:23:10.260 Samuel Roberts: What… what it, you know, what it does, it’s just… it’s just a system prompt, you know, for…
251 00:23:10.450 ⇒ 00:23:16.380 Samuel Roberts: For the API. It’s not, you know… so it’s… that one’s pretty standard, it’s not structured data coming back or anything, so…
252 00:23:16.940 ⇒ 00:23:21.230 Samuel Roberts: That’s definitely a decision we could go either way with, I think, and be probably fine.
253 00:23:22.250 ⇒ 00:23:22.920 Uttam Kumaran: Okay.
254 00:23:23.590 ⇒ 00:23:32.600 Samuel Roberts: If we do want to go with LyingFuse, though, it might mean changing a little bit how we… because it’s stored in Superbase already, so where we’re fetching things from would have to change, but that’s, you know, not a crazy lift, I don’t think.
255 00:23:34.330 ⇒ 00:23:34.910 Uttam Kumaran: Okay.
256 00:23:38.860 ⇒ 00:23:45.150 Mustafa Raja: Yeah, the SDK also has, ability to create launch from within.
257 00:23:45.620 ⇒ 00:23:46.000 Samuel Roberts: Okay.
258 00:23:46.000 ⇒ 00:23:47.420 Mustafa Raja: Yeah, would doable.
259 00:23:47.420 ⇒ 00:23:58.529 Samuel Roberts: So, I mean, there’s a couple ways to go with that. One is not to change anything, just make sure people know that they can make agents and tweak prompts on the forge, perhaps build a better flow for that.
260 00:23:58.530 ⇒ 00:24:08.669 Samuel Roberts: The other is move everything into LangFuse, and then just have people change the prompts on LangFuse, test on the Forge. The other is to kind of replicate a little bit of LangFuse’s UI into the Forge.
261 00:24:09.160 ⇒ 00:24:11.219 Samuel Roberts: For those agents, specifically.
262 00:24:11.370 ⇒ 00:24:14.159 Samuel Roberts: Because it’s just structured data. It’s not structured data, I mean, it’s just text.
263 00:24:14.160 ⇒ 00:24:17.860 Uttam Kumaran: I’m down to add everybody to link views, like, I don’t… that’s fine, you know.
264 00:24:17.860 ⇒ 00:24:18.470 Samuel Roberts: Okay, yeah.
265 00:24:18.490 ⇒ 00:24:20.240 Uttam Kumaran: But, yeah, I don’t mind that at all.
266 00:24:20.240 ⇒ 00:24:20.830 Samuel Roberts: Okay.
267 00:24:24.680 ⇒ 00:24:28.709 Samuel Roberts: Yeah, okay, I think that… that works then. We might just have to…
268 00:24:29.500 ⇒ 00:24:35.970 Uttam Kumaran: And not everyone, like, you see the panic, Good, Gabe can totally… Like, I think…
269 00:24:36.150 ⇒ 00:24:51.540 Uttam Kumaran: after more people get introduced, like, that is gonna… that’s sort of, like, the step one, and then they’ll be like, okay, great, I’m, like, good at editing prompts, but now I want to learn, like, I need to change the UI a little bit, and, like, that’s how we want to kind of drag people through the process. Sorry, drag is probably not a great word, but, like.
270 00:24:51.860 ⇒ 00:24:58.559 Uttam Kumaran: You want to pull people into, like, oh, I can actually… Edit, like, applications, you know?
271 00:24:58.560 ⇒ 00:24:58.930 Samuel Roberts: Right.
272 00:24:58.930 ⇒ 00:25:01.770 Uttam Kumaran: And I think that’s a really great first,
273 00:25:02.710 ⇒ 00:25:18.420 Uttam Kumaran: first, sort of, piece of that, when you’re… you’re… you learn, like, oh, I can make a tweak, test it, and then, like, I’m sure… I haven’t used language heavily, but there’s probably some, like, okay, now I’m gonna push this to production. Okay, but the prompt is, like, a big part of the product, you know?
274 00:25:18.420 ⇒ 00:25:29.090 Samuel Roberts: Yeah. No, I think so. I think you’re right. And LankFuse, if we’re good with just adding people to there, yeah, I don’t know how the billing and stuff works, but if it… if that’s good, the versioning there is definitely worth having, because we’d have to re…
275 00:25:29.450 ⇒ 00:25:35.059 Samuel Roberts: redo all… like, nothing is versioned in the Supabase stored agents right now, so that’s definitely worth having in link views.
276 00:25:36.000 ⇒ 00:25:37.150 Samuel Roberts: Cool. Okay.
277 00:25:38.250 ⇒ 00:25:42.050 Samuel Roberts: There’s not a ton of agents there right now, so we could probably move those over pretty easily.
278 00:25:42.850 ⇒ 00:25:46.089 Samuel Roberts: And restructure the way it’s fetching from Supabase.
279 00:25:48.000 ⇒ 00:25:53.709 Samuel Roberts: And I think they… yeah, we can do folders and stuff, we just couldn’t with N8N, is that right, Mustafa?
280 00:25:56.300 ⇒ 00:25:57.980 Mustafa Raja: Yes, that is correct.
281 00:25:57.980 ⇒ 00:26:07.389 Samuel Roberts: Okay, yeah, so LangFuse has organizational, just… things are all top-level right now, because N8in needed it that way, but we can… now that we’re moving things over, it’s fine. All right, cool. That sounds like a good plan.
282 00:26:08.140 ⇒ 00:26:09.750 Mustafa Raja: We can structure it better now.
283 00:26:10.050 ⇒ 00:26:11.380 Samuel Roberts: Yeah, definitely.
284 00:26:11.730 ⇒ 00:26:12.460 Samuel Roberts: Cool.
285 00:26:14.590 ⇒ 00:26:16.039 Uttam Kumaran: Cool. Yeah, go ahead.
286 00:26:16.040 ⇒ 00:26:17.929 Gabriel Lam: Just one last thing,
287 00:26:18.430 ⇒ 00:26:27.080 Gabriel Lam: I had a chat with Rico yesterday, and we were just bringing up how, in our new summary, we do have action items, that are based on, like, Brainforge.
288 00:26:27.490 ⇒ 00:26:30.290 Gabriel Lam: action items for Brainforge versus the client.
289 00:26:30.530 ⇒ 00:26:34.400 Gabriel Lam: And he brought up having… the…
290 00:26:35.580 ⇒ 00:26:41.200 Gabriel Lam: A different view of having, like, person-by-person action items, and wanting to make them…
291 00:26:41.770 ⇒ 00:26:44.810 Gabriel Lam: line up with the linear tickets, and so,
292 00:26:45.140 ⇒ 00:26:53.509 Gabriel Lam: Mustafa already took a stab at it to make sure that these, these, titles are being carried across.
293 00:26:54.180 ⇒ 00:26:59.520 Gabriel Lam: I just think that’s something good to make sure that… because initially they were running separately.
294 00:26:59.720 ⇒ 00:27:07.100 Gabriel Lam: back when it was on NAN, to make sure that we aren’t having hallucinations or duplicate information.
295 00:27:07.190 ⇒ 00:27:08.350 Uttam Kumaran: Okay.
296 00:27:08.350 ⇒ 00:27:11.970 Gabriel Lam: I don’t know how much progress has been made there, but I know Mustafa was able to…
297 00:27:13.510 ⇒ 00:27:19.099 Uttam Kumaran: Yeah, I mean, the one thing that, like, I think the theme is we’re going to produce
298 00:27:19.250 ⇒ 00:27:24.929 Uttam Kumaran: Videos, text, linear, and all of this we want to prioritize having, like.
299 00:27:25.330 ⇒ 00:27:28.350 Uttam Kumaran: deeper linking, I mean, deep linking between.
300 00:27:28.780 ⇒ 00:27:31.270 Uttam Kumaran: So…
301 00:27:31.580 ⇒ 00:27:49.539 Uttam Kumaran: I think that’s, like, what I… a great example of this is the stand-up assistant, where we have links to everything, like, really, really amazing. I think we will also start to, like, as much of that as we can do, it’s very helpful. Like, here’s not only the bullet, but the related ticket. That goes, like, one step beyond, because
302 00:27:49.630 ⇒ 00:28:08.000 Uttam Kumaran: any… everyone on the team here is gonna look at that and then be like, okay, where’s the ticket? Where’s the ticket? You know? So, one is the AI can find the relevant ticket. If not, it can… at that moment, you know, another opportunity is to basically say, okay, I know I need to create this ticket. So what you’re seeing, even in our… there are pieces of the ticket creation agent
303 00:28:08.000 ⇒ 00:28:12.439 Uttam Kumaran: That can be leveraged here, where, hey, here’s exactly, like, here’s the output.
304 00:28:12.630 ⇒ 00:28:25.410 Uttam Kumaran: here’s a summary. Oh, I realize that there’s no tickets for some of these, I should go to create them. So this is where, like, kind of all the small pieces of the various systems can get leveraged elsewhere. Like, there is a…
305 00:28:25.410 ⇒ 00:28:35.390 Uttam Kumaran: ticket generation endpoint, right? Or, like, hey, call linear, compare the results, create tickets. So these pieces are all modular, and we want to use them
306 00:28:35.490 ⇒ 00:28:37.910 Uttam Kumaran: Everywhere, and have that deep linking, you know?
307 00:28:38.560 ⇒ 00:28:45.109 Samuel Roberts: Yeah, we just added the, every linear ticket should automatically have a link to the meeting on the forge that it was from, as well.
308 00:28:45.240 ⇒ 00:28:56.119 Samuel Roberts: Just like you can have things on Slack and things, so that is now… meeting source is there, because I figured that was exactly something we were going to want, keeping… to keep track of,
309 00:28:56.670 ⇒ 00:29:01.559 Samuel Roberts: I think in the description, sometimes it has the timestamps and stuff, but we can eventually add more of that too, so…
310 00:29:02.680 ⇒ 00:29:05.349 Samuel Roberts: But yeah, that’s definitely something on top of mind.
311 00:29:09.460 ⇒ 00:29:10.180 Uttam Kumaran: Okay.
312 00:29:12.680 ⇒ 00:29:13.470 Uttam Kumaran: Cool.
313 00:29:13.590 ⇒ 00:29:16.799 Uttam Kumaran: Well, I think that’s all I had. I’m gonna spend the next…
314 00:29:17.500 ⇒ 00:29:25.750 Uttam Kumaran: hour, two hours just reviewing docs. I’m gonna review this… the Snowflake dbt doc, I’ll review ACE, put some notes, in…
315 00:29:26.760 ⇒ 00:29:31.240 Uttam Kumaran: text SQL, I’ll review, the ABC migration plan.
316 00:29:31.450 ⇒ 00:29:33.899 Uttam Kumaran: I know I need to do some stuff for…
317 00:29:34.180 ⇒ 00:29:40.029 Uttam Kumaran: DTA, just ping me, I’m just gonna kind of focus on, like, reviewing as much stuff as possible in the next, like, hour or so.
318 00:29:40.230 ⇒ 00:29:43.080 Uttam Kumaran: But yeah.
319 00:29:43.830 ⇒ 00:29:46.889 Uttam Kumaran: And then, I’ll probably just listen to the next meeting.
320 00:29:50.440 ⇒ 00:29:59.469 Uttam Kumaran: Okay, cool, and then, yeah, I’m gonna be off next 2 days. I’ll probably… I’ll be on today, probably until, like, afternoon, and then I’ll be on my phone, but,
321 00:29:59.700 ⇒ 00:30:01.929 Uttam Kumaran: Of course, just, like, pygmy.
322 00:30:02.150 ⇒ 00:30:11.700 Uttam Kumaran: And then if you need to call me, I’ll check Slack, like, in the mornings and stuff tomorrow. And then for folks that are still staying on, yeah, if there is client work.
323 00:30:11.770 ⇒ 00:30:22.630 Uttam Kumaran: You know, feel free to take a stab at it. But it could be a good opportunity to think about tech debt items, could also be a good opportunity to work on internal stuff.
324 00:30:22.840 ⇒ 00:30:25.830 Uttam Kumaran: If you need ideas, let me know.
325 00:30:26.020 ⇒ 00:30:30.719 Uttam Kumaran: I think the one thing, Demolade, for context, Kate, I asked,
326 00:30:30.880 ⇒ 00:30:38.170 Uttam Kumaran: Mustafa and Ashwin need to work on a DBT setup, like, guide, and a Snowflake setup guide.
327 00:30:39.530 ⇒ 00:30:44.730 Uttam Kumaran: So, I think I’m gonna… I’m having them take the, kind of, the first stab at it, and then a few of us can review.
328 00:30:44.880 ⇒ 00:30:55.370 Uttam Kumaran: So that way, what we’re… what we’re hoping to do is just, one, increase the amount of people that can execute those sort of tool setups, as well as,
329 00:30:56.530 ⇒ 00:31:00.189 Uttam Kumaran: beat it up for the folks that are good at it, and get it right on the first time, so…
330 00:31:00.510 ⇒ 00:31:08.920 Uttam Kumaran: Yeah, I feel like we should pro- we’ll probably end up doing Snowflake, dbt, BigQuery… Polyatomic…
331 00:31:09.240 ⇒ 00:31:11.780 Uttam Kumaran: Like, the usual stuff that we do, so…
332 00:31:12.100 ⇒ 00:31:15.329 Samuel Roberts: I know I also mentioned, like, scripting or some CLI tools or stuff like that.
333 00:31:15.330 ⇒ 00:31:15.900 Uttam Kumaran: That’s it. Yeah.
334 00:31:15.900 ⇒ 00:31:21.510 Samuel Roberts: Like, maybe once that plan is written, like, loop me in, and I can try to help out if there’s things that can get automated.
335 00:31:21.510 ⇒ 00:31:22.090 Uttam Kumaran: Cool.
336 00:31:22.090 ⇒ 00:31:23.210 Samuel Roberts: Little more.
337 00:31:23.210 ⇒ 00:31:23.750 Uttam Kumaran: Cool.
338 00:31:24.050 ⇒ 00:31:24.860 Demilade Agboola: Descriptive.
339 00:31:25.080 ⇒ 00:31:25.460 Samuel Roberts: Or anything.
340 00:31:25.460 ⇒ 00:31:33.069 Demilade Agboola: I think… I think that would be very helpful, especially, like, if we have clients come in, and it’s just like, we need to initialize dbt for XYZ.
341 00:31:33.200 ⇒ 00:31:38.129 Demilade Agboola: You can just kind of go through the steps, like, an agent that can go through the steps and do it.
342 00:31:38.390 ⇒ 00:31:41.370 Demilade Agboola: And maybe just give prompts, okay, so what name do you want to give it?
343 00:31:41.370 ⇒ 00:31:42.260 Samuel Roberts: But, yeah.
344 00:31:42.490 ⇒ 00:31:57.050 Samuel Roberts: That’s exactly what I’m picturing, because I think most of these tools have CLI tools that we can tap into, and so if we can just prompt for a few names, a few keys, and it can probably do, you know, a good chunk of that stuff if it’s pretty well structured already.
345 00:31:57.390 ⇒ 00:31:59.040 Demilade Agboola: Amazing. Sounds great.
346 00:31:59.440 ⇒ 00:32:00.420 Samuel Roberts: Perfect, cool.
347 00:32:04.410 ⇒ 00:32:05.840 Uttam Kumaran: Okay, great.
348 00:32:06.000 ⇒ 00:32:07.330 Uttam Kumaran: Alright, thank you guys.
349 00:32:07.560 ⇒ 00:32:08.270 Samuel Roberts: Alrighty.
350 00:32:09.490 ⇒ 00:32:09.810 Uttam Kumaran: soon.
351 00:32:09.890 ⇒ 00:32:14.819 Demilade Agboola: What’s on? Yeah. Quick thing. So Urban Stems has…
352 00:32:15.180 ⇒ 00:32:23.599 Demilade Agboola: quick data issue, but again, it’s one of those things where it’s a looker issue versus a marked issue. Like, the data isn’t marked, they just can’t seem to see
353 00:32:23.910 ⇒ 00:32:32.740 Demilade Agboola: the inventory for… everyone else can drop, like, this is just… it’s an Urban STEM-specific thing, so you don’t have to be held host.
354 00:32:32.740 ⇒ 00:32:33.590 Samuel Roberts: I was thinking.
355 00:32:33.590 ⇒ 00:32:35.050 Uttam Kumaran: Okay.
356 00:32:35.050 ⇒ 00:32:47.460 Samuel Roberts: I might say, I’m off at, like, 2 today, and then for the next week, so if you need me, I probably may not have my notifications on on Slack next week, but you can push things through, I think, so… if you need me to get ahold of me.
357 00:32:47.890 ⇒ 00:32:54.460 Samuel Roberts: But yeah, otherwise, everyone have a good, good few days and a week, if I don’t talk to you.
358 00:32:54.690 ⇒ 00:32:58.499 Samuel Roberts: Thanks. AI team, I’ll probably talk to you before I get off, so we’re playing for the next few.
359 00:32:59.310 ⇒ 00:33:01.340 Samuel Roberts: Few days, at least. Cool. Alright.
360 00:33:01.620 ⇒ 00:33:02.359 Samuel Roberts: See y’all.
361 00:33:03.000 ⇒ 00:33:04.590 Demilade Agboola: Have a great Thanksgiving.
362 00:33:04.590 ⇒ 00:33:05.280 Gabriel Lam: Yeah, thank you.
363 00:33:06.680 ⇒ 00:33:07.570 Mustafa Raja: Thank you.
364 00:33:10.620 ⇒ 00:33:18.759 Uttam Kumaran: Yeah, so it’s a looker thing. So, again, so this is… this is kind of like the… you’re tagged in the conversation, so you can actually kind of see it. It’s going on right now.
365 00:33:18.760 ⇒ 00:33:26.109 Demilade Agboola: Okay, yeah, I’ll see, I’ll take a look right now. It’s basically the same thing I’ve been saying. We have great data, the looker doesn’t allow us to shine.
366 00:33:27.060 ⇒ 00:33:29.170 Demilade Agboola: It’s one of those things.
367 00:33:29.170 ⇒ 00:33:41.470 Uttam Kumaran: Yeah, I… I told this to… I told this to, so, Zach… You know?
368 00:33:41.580 ⇒ 00:33:49.940 Uttam Kumaran: And… I can keep sending this to him, but…
369 00:33:55.720 ⇒ 00:34:05.779 Demilade Agboola: Because I know when the issue came up today, I was like, okay, I haven’t changed any logic with available for sale by Fulfillment center, so there should be no reason why
370 00:34:06.500 ⇒ 00:34:11.779 Demilade Agboola: like, we’re not able to see, like, Baltimore, how many items are available for sale there.
371 00:34:12.070 ⇒ 00:34:14.770 Uttam Kumaran: And so, when… Okay, so…
372 00:34:15.370 ⇒ 00:34:20.520 Uttam Kumaran: So, let me give you… let me… I didn’t know if I went into this yesterday, so basically, Zach…
373 00:34:26.750 ⇒ 00:34:29.770 Uttam Kumaran: Long story short, and like, you can just keep this…
374 00:34:29.929 ⇒ 00:34:36.200 Uttam Kumaran: with us, but Zach is just trying to find out, like, whether they have the right resources on the data team.
375 00:34:37.889 ⇒ 00:34:41.620 Uttam Kumaran: Like, and you can, like, read kind of between the lines on that one.
376 00:34:42.080 ⇒ 00:34:47.270 Uttam Kumaran: He’s just trying to find out exactly, because he knows that we’ve been doing a lot of work.
377 00:34:47.600 ⇒ 00:34:50.440 Uttam Kumaran: But… and he’s aware of the fact that
378 00:34:50.719 ⇒ 00:34:57.509 Uttam Kumaran: between the analysts and Emily, he’s aware of, like, hey, maybe these folks are just not, like.
379 00:34:58.400 ⇒ 00:35:07.089 Uttam Kumaran: technical or talented enough to take some of our stuff and move it forward, and he’s like, I don’t know the answer.
380 00:35:08.460 ⇒ 00:35:17.700 Uttam Kumaran: And he’s like, I need… I wanted to… he basically is like, I need you guys to step back so I can find out the answer of what can these folks
381 00:35:18.490 ⇒ 00:35:20.750 Uttam Kumaran: Do, now that you’ve finished the marts.
382 00:35:22.050 ⇒ 00:35:23.130 Demilade Agboola: Okay. And…
383 00:35:23.130 ⇒ 00:35:25.350 Uttam Kumaran: Whether we have the right people or not.
384 00:35:26.360 ⇒ 00:35:29.759 Uttam Kumaran: So… I… this is where, like.
385 00:35:29.910 ⇒ 00:35:41.779 Uttam Kumaran: it’s internal politics. Like, the thing is, Zach, they’re spending a lot of money on the data team, like, analysts, Emily, and then us, and he’s like, we just don’t have the budget
386 00:35:42.240 ⇒ 00:35:49.399 Uttam Kumaran: So, he’s like, you guys won’t go anywhere, but I need to understand, like, what the true budget for this team needs to be.
387 00:35:49.570 ⇒ 00:35:54.250 Uttam Kumaran: And, like, whether we have the right people or not. So, that’s the context for our next…
388 00:35:54.730 ⇒ 00:35:58.329 Uttam Kumaran: phase with them, is that we’re just gonna kind of be on, like, help mode.
389 00:35:58.880 ⇒ 00:36:00.990 Demilade Agboola: It’s… it’s… I…
390 00:36:01.150 ⇒ 00:36:07.939 Uttam Kumaran: I was like, alright, like… I said, like, dude, that’s kind of hard, because…
391 00:36:08.430 ⇒ 00:36:11.329 Uttam Kumaran: Where, like, it’s gonna be hard for me to watch.
392 00:36:11.620 ⇒ 00:36:16.990 Uttam Kumaran: But he’s like, I honestly kind of need things to just… Bail and see what happens.
393 00:36:17.600 ⇒ 00:36:19.600 Uttam Kumaran: Versus you guys picking everything up.
394 00:36:20.420 ⇒ 00:36:23.280 Uttam Kumaran: Because we have to build, like, the internal muscle for this.
395 00:36:23.400 ⇒ 00:36:26.089 Uttam Kumaran: So, that being said, on tickets like this.
396 00:36:26.300 ⇒ 00:36:30.200 Uttam Kumaran: As long as we have our end covered, like…
397 00:36:31.640 ⇒ 00:36:33.709 Uttam Kumaran: I think we… we pass it to them.
398 00:36:35.840 ⇒ 00:36:46.249 Uttam Kumaran: we passed it to Emily, because we… we weren’t, like, on… we… we… we did our best to try to get this far, this fast, and there’s still more stuff to do, and we’re…
399 00:36:46.490 ⇒ 00:36:52.900 Uttam Kumaran: we just haven’t been requested to do so, so… I don’t… like, Emily isn’t really privy to this.
400 00:36:53.720 ⇒ 00:36:59.730 Uttam Kumaran: And so, for us, like, our next contract with them is gonna be purely Hourly.
401 00:37:00.020 ⇒ 00:37:01.810 Uttam Kumaran: But for the most part, I think it’ll…
402 00:37:03.410 ⇒ 00:37:07.700 Uttam Kumaran: it may just be us and Slack, like, handling things, like, as we need to, you know?
403 00:37:09.780 ⇒ 00:37:12.319 Demilade Agboola: I have, I have that.
404 00:37:12.840 ⇒ 00:37:17.560 Demilade Agboola: But is there a scope of what we handle, though? Because even though it’s an hourly…
405 00:37:18.050 ⇒ 00:37:22.770 Demilade Agboola: Contracts. Is there scope to what we should handle, or what’s out of scope?
406 00:37:24.290 ⇒ 00:37:32.089 Uttam Kumaran: He basically said just, like, to assist on, like.
407 00:37:32.340 ⇒ 00:37:36.679 Uttam Kumaran: Think… when things are, like, really failing, or to be there as, like, a backstop.
408 00:37:38.980 ⇒ 00:37:45.500 Uttam Kumaran: Most likely, what will happen is, as new asks come in, I will run it by Zach to get approval.
409 00:37:45.730 ⇒ 00:37:47.020 Uttam Kumaran: For us to work on it.
410 00:37:47.720 ⇒ 00:37:54.549 Uttam Kumaran: But we will… we… we won’t… we won’t be nearly, the amount of hours that we were before.
411 00:37:55.240 ⇒ 00:37:56.129 Uttam Kumaran: Go ahead.
412 00:37:56.370 ⇒ 00:38:00.759 Uttam Kumaran: So he’s try… he’s trying to or… he’s basically trying to orchestrate, like.
413 00:38:00.980 ⇒ 00:38:05.360 Uttam Kumaran: A little bit of a safe, Explosion.
414 00:38:06.870 ⇒ 00:38:13.390 Uttam Kumaran: just to see, like, okay, what can the… what can our folks actually do?
415 00:38:15.810 ⇒ 00:38:20.990 Demilade Agboola: Which… I think, if we’re going to negotiate another contract with them.
416 00:38:20.990 ⇒ 00:38:25.250 Uttam Kumaran: I tried to get him to do a minimum. Like, I tried to say, hey, do 5K minimum.
417 00:38:25.850 ⇒ 00:38:28.360 Uttam Kumaran: And he was like, we don’t even have budget for that.
418 00:38:28.830 ⇒ 00:38:31.210 Uttam Kumaran: And I was like, okay.
419 00:38:31.530 ⇒ 00:38:43.989 Uttam Kumaran: like, what the… what the F? And he’s like, yeah, I mean, he just said, like, look, we have all these data people, and nobody’s not… people aren’t doing stuff, we’re doing, like, budgeting and spreadsheets, he’s just… I think he’s just concerned that they don’t have the talent, you know?
420 00:38:44.710 ⇒ 00:38:53.529 Demilade Agboola: to be honest, I think what they have is a talent distribution problem. They have too many analysts, very few AEs, and the analysts
421 00:38:53.700 ⇒ 00:38:55.330 Demilade Agboola: Like, the problem is…
422 00:38:55.330 ⇒ 00:39:02.520 Uttam Kumaran: They’re not that technical, like, they… they’re… yeah, they can totally handle, like, a bunch more scope, dude, like, each of them, you know?
423 00:39:02.810 ⇒ 00:39:12.440 Demilade Agboola: But I also feel like it’s one of those things where it’s too downstream heavy, very little upstream, and the people downstream also aren’t necessarily always the best, because
424 00:39:13.130 ⇒ 00:39:19.639 Demilade Agboola: like, for instance, this available for sale issue, I don’t know how long it’s been on, but just think about the fact that, like.
425 00:39:19.780 ⇒ 00:39:27.290 Demilade Agboola: You have a company that sells, like, that delivers flowers nationwide, and you have
426 00:39:27.870 ⇒ 00:39:46.430 Demilade Agboola: eventually that you have not been seeing for a minute. That’s actually crazy to think about, actually. And it’s literally because the data is available, but the setup in Looker isn’t done accurately, or isn’t done well enough. It could be a bad join, could be an unintentional filter or something, and all of a sudden.
427 00:39:46.860 ⇒ 00:39:50.190 Demilade Agboola: You’re not seeing, like, a ton of your data, or a bunch of your data.
428 00:39:50.400 ⇒ 00:39:56.810 Demilade Agboola: So yeah, I feel like it’s those little things that are, like, red flags, and I have been trying to point out with, like, the lookup problem.
429 00:39:56.920 ⇒ 00:40:06.840 Demilade Agboola: that we do have the data, we’re giving them the data, but once it gets into Looker, it’s its own black box, and, like, what comes out isn’t necessarily what we can control.
430 00:40:07.860 ⇒ 00:40:09.210 Uttam Kumaran: Hmm. Okay.
431 00:40:09.810 ⇒ 00:40:18.370 Demilade Agboola: So for the next scope, if we do get a next scope, we should definitely try and get someone who, on our team, can be responsible for, like, the lookout aspect of it.
432 00:40:19.520 ⇒ 00:40:20.889 Uttam Kumaran: Yeah, yeah, yeah, okay.
433 00:40:20.920 ⇒ 00:40:24.900 Demilade Agboola: That’s what the, you know, stakeholders see and will interact with.
434 00:40:26.490 ⇒ 00:40:30.680 Uttam Kumaran: Yeah, so for the next scope, we are planning on… okay, I see what you mean.
435 00:40:31.970 ⇒ 00:40:51.159 Demilade Agboola: Yeah, so if we do get no contractions, yeah, great, but we need to just have budget for someone who will be in look at specifically. Okay. Because we don’t want to have that same thing of, like, oh, we push out all this data, someone then does an explore that is a bad join, and it knocks off part of the data.
436 00:40:51.230 ⇒ 00:40:58.719 Demilade Agboola: And now, stakeholders are like, this number feels lower than it should be, or, you know, this number isn’t how we expect things to be.
437 00:40:58.990 ⇒ 00:41:03.029 Demilade Agboola: So I think that’s a huge, like, part of it as well.
438 00:41:03.860 ⇒ 00:41:04.480 Uttam Kumaran: Okay.
439 00:41:05.620 ⇒ 00:41:08.330 Demilade Agboola: Okay, weren’t we able to, like…
440 00:41:09.560 ⇒ 00:41:12.610 Demilade Agboola: Like, get them off of building in the looker?
441 00:41:14.700 ⇒ 00:41:18.520 Uttam Kumaran: Yeah, I mean, you could totally do that, but, like, you have to have a control in Looker.
442 00:41:18.870 ⇒ 00:41:22.079 Uttam Kumaran: There’s just so many analysts that are building stuff in there, you know?
443 00:41:22.840 ⇒ 00:41:30.870 Awaish Kumar: No, I mean, like, when we have dbt, where we can do summary tables, then again, you go into Looker, you do development.
444 00:41:31.130 ⇒ 00:41:40.519 Uttam Kumaran: No, no, no, I know, it’s… but that’s, like, it’s… what Demi is saying is that’s not in our scope to manage Looker right now, and that’s, like, what he’s telling me to…
445 00:41:41.290 ⇒ 00:41:43.029 Uttam Kumaran: Go get them to agree to.
446 00:41:43.400 ⇒ 00:41:48.519 Awaish Kumar: Yeah, I understand, I’m just curious about, like, why having both the setups, right?
447 00:41:48.900 ⇒ 00:41:55.710 Uttam Kumaran: Because they don’t know, they have no clue, none of their team is trained on dbt, like, they’re just not familiar with this, they’re not technical.
448 00:41:56.490 ⇒ 00:41:57.489 Demilade Agboola: Yeah, and they do a lot.
449 00:41:57.490 ⇒ 00:42:03.070 Uttam Kumaran: And they’re all running, like, individually. There’s no central, like, data governance team.
450 00:42:03.480 ⇒ 00:42:08.259 Uttam Kumaran: Right. And so, you make a purchase of a tool like Looker, and you make it open for everybody.
451 00:42:08.690 ⇒ 00:42:14.670 Uttam Kumaran: And you don’t prioritize, like, a team, you don’t give us enough budget to, like, do those types of things, like…
452 00:42:15.280 ⇒ 00:42:16.090 Uttam Kumaran: You know?
453 00:42:19.840 ⇒ 00:42:20.869 Demilade Agboola: It’s just, it’s just…
454 00:42:21.050 ⇒ 00:42:31.740 Demilade Agboola: I just feel like it was a team set up by people who just feel like, we need analysts, we need analysts, we need analysts, without trying to think of how to… like, what do the analysts actually do?
455 00:42:32.550 ⇒ 00:42:43.110 Demilade Agboola: And how do they have the infrastructure for that? So, I’m always never really a fan of, like, Looker, like, being so Looker heavy, because Looker itself relies on, like, a lot of logic.
456 00:42:43.380 ⇒ 00:42:52.120 Demilade Agboola: And if you have dbt logic, it means that you’re… you have logic in two places, and that’s a huge problem eventually, because once things start to break.
457 00:42:52.190 ⇒ 00:43:05.970 Demilade Agboola: you have to start troubleshooting Looker, and then troubleshooting, dbt, versus, like, a Tableau or, like, you know, whatever other tool, where you’re not doing so much within Tableau, so if things are breaking, it’s probably dbt.
458 00:43:06.270 ⇒ 00:43:08.190 Demilade Agboola: That kind of thing.
459 00:43:09.170 ⇒ 00:43:15.560 Demilade Agboola: But yeah, I think Luca is doing too much, there are too many people, and it’s… it’s harder to keep track of things when things are breaking.
460 00:43:16.450 ⇒ 00:43:16.990 Uttam Kumaran: Yeah.
461 00:43:24.530 ⇒ 00:43:25.310 Uttam Kumaran: Okay.
462 00:43:27.350 ⇒ 00:43:28.210 Demilade Agboola: Alright.
463 00:43:28.530 ⇒ 00:43:29.270 Demilade Agboola: Alright.
464 00:43:29.270 ⇒ 00:43:29.979 Uttam Kumaran: That’s all I had.
465 00:43:30.550 ⇒ 00:43:31.240 Demilade Agboola: Same here.
466 00:43:31.730 ⇒ 00:43:32.330 Uttam Kumaran: Okay.
467 00:43:32.570 ⇒ 00:43:35.060 Uttam Kumaran: Awash, maybe I can… you can stay on and can just chat.
468 00:43:35.810 ⇒ 00:43:36.460 Awaish Kumar: Okay.
469 00:43:37.740 ⇒ 00:43:39.859 Uttam Kumaran: How did it go? How’d the interview go?
470 00:43:41.970 ⇒ 00:43:52.840 Awaish Kumar: Yeah, it went fine, like, but he… he mentioned he’s an entry-level person. As an entry-level person, I think he was good, but I see a few things, like,
471 00:43:53.030 ⇒ 00:44:09.039 Awaish Kumar: thinking, like, when you are given the new requirements, how would you approach it? Like, Google search, or use of AI tools, or come up with industry standards? Like, that was missing, I couldn’t…
472 00:44:09.380 ⇒ 00:44:15.389 Awaish Kumar: like, get that response from him, but after, like, I provided some…
473 00:44:15.910 ⇒ 00:44:19.290 Awaish Kumar: Responses in the interview to… to engage it, like them…
474 00:44:19.470 ⇒ 00:44:25.960 Awaish Kumar: at an entry level, I see, like, he can… Learn, but yeah, not…
475 00:44:26.900 ⇒ 00:44:31.190 Awaish Kumar: Not a… not a senior level person which can just go and do stuff.
476 00:44:31.700 ⇒ 00:44:32.290 Uttam Kumaran: Okay.
477 00:44:35.130 ⇒ 00:44:36.510 Uttam Kumaran: Okay, makes sense.
478 00:44:38.850 ⇒ 00:44:43.519 Uttam Kumaran: Cool. Yeah, what else from, from your side? Anything for today?
479 00:44:45.200 ⇒ 00:44:57.409 Awaish Kumar: Like, I shared some tables for, like, Hedra. She was asking for ARR last time, so I created all these daily snapshot tables, but then that is confusing.
480 00:44:57.570 ⇒ 00:45:06.579 Awaish Kumar: Like, if she just sums it, like, it will result in bad numbers, so I created her summary tables for her, so she don’t have to do it in a query.
481 00:45:08.430 ⇒ 00:45:12.650 Awaish Kumar: then I try to, yeah, I ask for…
482 00:45:13.220 ⇒ 00:45:20.930 Awaish Kumar: SLA for ingestion from Polytomic on rest of the data, because that is the, basically.
483 00:45:21.330 ⇒ 00:45:37.830 Awaish Kumar: problem. I… when I, like, when she shared the numbers in the beginning, like, the numbers are negative or things like that, that’s only because there’s not enough data. So we have… VIN-wise, you got a discounted,
484 00:45:38.000 ⇒ 00:45:41.569 Awaish Kumar: Item, but you don’t get the item which has actual
485 00:45:41.830 ⇒ 00:45:45.160 Awaish Kumar: amount, so it resulted in negative values.
486 00:45:45.380 ⇒ 00:45:46.600 Uttam Kumaran: Okay. So.
487 00:45:46.600 ⇒ 00:45:56.069 Awaish Kumar: I took a different approach because of that. We might re… have to re… like, have another pass when we have all the data.
488 00:45:56.860 ⇒ 00:45:57.420 Uttam Kumaran: Okay.
489 00:45:58.990 ⇒ 00:45:59.950 Uttam Kumaran: Okay, that’s fine.
490 00:45:59.950 ⇒ 00:46:00.540 Awaish Kumar: foot.
491 00:46:00.760 ⇒ 00:46:10.860 Awaish Kumar: Yeah, for Hedra, I think there’s nothing else, right? For now. She mentioned a few things, but that’s not possible until we have the rest of the data.
492 00:46:12.030 ⇒ 00:46:12.580 Uttam Kumaran: Okay.
493 00:46:15.350 ⇒ 00:46:18.730 Awaish Kumar: So, I don’t know if I’m not anything, then I…
494 00:46:19.610 ⇒ 00:46:31.390 Awaish Kumar: Yeah, Ash… yeah, I met with Ashwini and Zoran, today in the… during the day. I shared the content, like, context for agent tickets and,
495 00:46:31.950 ⇒ 00:46:35.340 Awaish Kumar: I think we finalized the catalog’s thing as well, and, you know…
496 00:46:35.340 ⇒ 00:46:35.950 Uttam Kumaran: Okay.
497 00:46:38.710 ⇒ 00:46:41.150 Awaish Kumar: Yeah, that’s all.
498 00:46:41.650 ⇒ 00:46:46.179 Awaish Kumar: messaging me regarding some tickets on Adrian, I will look into it, and see if I…
499 00:46:46.330 ⇒ 00:46:57.359 Awaish Kumar: have to work on anything, but otherwise, there’s nothing else there as well. So, for Eden, for urban stems, I don’t know if we are taking on anything else, or not.
500 00:46:57.360 ⇒ 00:46:59.409 Uttam Kumaran: That’s fine. Yeah, we can leave Urban Stems.
501 00:46:59.580 ⇒ 00:47:01.840 Uttam Kumaran: Maybe I,
502 00:47:02.130 ⇒ 00:47:07.300 Uttam Kumaran: I may ask Awayish, if you have time today, you want to work on, like, a SOP for…
503 00:47:08.090 ⇒ 00:47:10.450 Uttam Kumaran: dbt and Snowflake setup.
504 00:47:12.230 ⇒ 00:47:16.019 Awaish Kumar: Okay, I… yeah, I have that, and I can go over.
505 00:47:16.140 ⇒ 00:47:22.660 Awaish Kumar: So I, like, we created one few, like, in a meeting, I think, but, like, do you want me to…
506 00:47:24.310 ⇒ 00:47:25.010 Awaish Kumar: lose that.
507 00:47:25.010 ⇒ 00:47:25.840 Uttam Kumaran: Yeah.
508 00:47:26.220 ⇒ 00:47:29.840 Uttam Kumaran: I, I don’t know, what do you think? Like, well, kind of like…
509 00:47:30.500 ⇒ 00:47:35.509 Uttam Kumaran: I was… I wanted to have an SOP, and then I was gonna have Sam build, like, a…
510 00:47:35.800 ⇒ 00:47:37.530 Uttam Kumaran: script, like a CLI script.
511 00:47:38.280 ⇒ 00:47:41.719 Uttam Kumaran: to connect to Snowflake CLI or dbt CLI eventually.
512 00:47:41.820 ⇒ 00:47:47.299 Uttam Kumaran: I was wondering if you could… if you could write the steps, though, and then…
513 00:47:48.120 ⇒ 00:47:55.530 Uttam Kumaran: Basically, what we wanted… what we want to do is, like, anybody in the company should be able to, like, initialize Snowflake based on this doc, you know?
514 00:47:56.800 ⇒ 00:48:03.450 Awaish Kumar: Yeah, like… Okay, so for Snowflake, it…
515 00:48:03.690 ⇒ 00:48:20.390 Awaish Kumar: there is, like, I see there is a requirement, like, SOP is… having SOP is good, because it has to run, create all users, warehouses, and things like that. For dbt, I don’t know what… like, if it is not dbt Cloud, what do we need? Like, we just need to install.
516 00:48:21.030 ⇒ 00:48:23.469 Uttam Kumaran: Well, you have to set up the… you’ll have to,
517 00:48:24.410 ⇒ 00:48:30.220 Uttam Kumaran: set up, well, if it is dbt… well, one, we do have to have the one for dbt Cloud.
518 00:48:30.580 ⇒ 00:48:36.809 Uttam Kumaran: For core, you do have to set up… you do have to confirm the warehouse credentials.
519 00:48:36.930 ⇒ 00:48:42.050 Uttam Kumaran: You have to load it into your profiles to create the end file, break the…
520 00:48:42.280 ⇒ 00:48:44.490 Uttam Kumaran: Break the repo, of course, like…
521 00:48:44.670 ⇒ 00:48:50.419 Uttam Kumaran: it’s actually… those are the things that me and you would, like, we would just gloss over, but, like, that’s… I want to have almost, like.
522 00:48:50.710 ⇒ 00:48:53.240 Uttam Kumaran: dummy-proof.
523 00:48:53.470 ⇒ 00:48:57.079 Awaish Kumar: You just want to have a… steps, like, if you…
524 00:48:57.080 ⇒ 00:48:57.430 Uttam Kumaran: Yeah.
525 00:48:57.430 ⇒ 00:49:01.530 Awaish Kumar: There’s no repo, there’s nothing. How to start? Okay.
526 00:49:01.530 ⇒ 00:49:07.370 Uttam Kumaran: Yeah, and then also, like, for example, in your dbt project file, you have to change it to the client name.
527 00:49:07.800 ⇒ 00:49:17.640 Uttam Kumaran: And how to, like… for example, we eventually want to get to the end of the dbt thing, so to show that you can run dbt build.
528 00:49:17.750 ⇒ 00:49:23.950 Uttam Kumaran: And the two test models are present in your warehouse, right? The two example models.
529 00:49:24.190 ⇒ 00:49:27.910 Awaish Kumar: So, ideally, all… we can write it in Notion, and then it can end up.
530 00:49:27.910 ⇒ 00:49:34.629 Uttam Kumaran: this can end up in the repo, in the template repo as well. So we throw all… as much of this in the template repo, I think.
531 00:49:36.080 ⇒ 00:49:36.590 Uttam Kumaran: And then…
532 00:49:36.590 ⇒ 00:49:38.439 Awaish Kumar: Yeah, maybe I can also come up.
533 00:49:38.440 ⇒ 00:49:46.029 Uttam Kumaran: that he can actually… he can actually do this, like, he can do this as a… as a CL… he can do this in the CLI, so, eventually.
534 00:49:47.980 ⇒ 00:49:52.240 Uttam Kumaran: Like, so he can write a bash script or something that executes all this, and then that’ll be perfect.
535 00:49:55.690 ⇒ 00:49:56.460 Awaish Kumar: Okay.
536 00:49:56.720 ⇒ 00:49:59.179 Awaish Kumar: Okay, I can write the steps, and…
537 00:50:00.130 ⇒ 00:50:02.179 Awaish Kumar: Maybe I can also get to the script.
538 00:50:02.370 ⇒ 00:50:03.650 Awaish Kumar: Yeah.
539 00:50:05.250 ⇒ 00:50:07.660 Uttam Kumaran: So I’ll just send these to you,
540 00:50:13.320 ⇒ 00:50:18.370 Uttam Kumaran: And then, you know we have that doc that’s, like, how to set up dbt, how to… how do we structure dbt?
541 00:50:18.820 ⇒ 00:50:19.340 Uttam Kumaran: So, like.
542 00:50:19.340 ⇒ 00:50:19.890 Awaish Kumar: Yeah.
543 00:50:20.410 ⇒ 00:50:23.469 Uttam Kumaran: We can link out to that, but these are both of the… both of the ones.
544 00:50:24.890 ⇒ 00:50:28.689 Uttam Kumaran: I was gonna work on this later, but if you want to take a stab at it, that’s great.
545 00:50:31.540 ⇒ 00:50:32.340 Awaish Kumar: Okay.
546 00:50:33.430 ⇒ 00:50:37.410 Awaish Kumar: how, yeah, like… So, like, if somebody…
547 00:50:39.360 ⇒ 00:50:43.360 Awaish Kumar: creates a bash script. That means that script needs to be…
548 00:50:44.150 ⇒ 00:50:48.410 Awaish Kumar: Sent over to the user’s computer so they can run it.
549 00:50:50.610 ⇒ 00:50:51.830 Uttam Kumaran: Okay, so…
550 00:50:51.830 ⇒ 00:50:58.120 Awaish Kumar: What if we create some… what if we create some repository which is installable?
551 00:50:58.740 ⇒ 00:50:59.700 Awaish Kumar: Right?
552 00:50:59.870 ⇒ 00:51:01.679 Awaish Kumar: On the weekend, then…
553 00:51:02.190 ⇒ 00:51:09.800 Awaish Kumar: use that in our, like, in our projects, like, pip install this repo, and then you can directly run commands.
554 00:51:10.000 ⇒ 00:51:15.099 Awaish Kumar: to create… Like, the structure and everything, and then we can.
555 00:51:15.100 ⇒ 00:51:15.450 Uttam Kumaran: Yeah.
556 00:51:15.450 ⇒ 00:51:18.469 Awaish Kumar: Adding on stuff into that repo.
557 00:51:18.780 ⇒ 00:51:19.320 Uttam Kumaran: Yeah.
558 00:51:21.520 ⇒ 00:51:24.970 Awaish Kumar: So it will be kind of packaged, which you have to run, you don’t have to share.
559 00:51:24.970 ⇒ 00:51:28.210 Uttam Kumaran: The thing is, not every… not every client will have, like.
560 00:51:28.580 ⇒ 00:51:32.820 Uttam Kumaran: some client will be BigQuery, some client will be… so don’t over-engineer for now.
561 00:51:34.320 ⇒ 00:51:37.469 Uttam Kumaran: You know, I’m not as concerned with this happening
562 00:51:37.710 ⇒ 00:51:41.039 Uttam Kumaran: I’m like, if we can get this set up to under 30 minutes.
563 00:51:41.470 ⇒ 00:51:44.100 Uttam Kumaran: Getting it from 30 to 5 minutes, I’m not, like, huh?
564 00:51:45.530 ⇒ 00:51:49.469 Awaish Kumar: Yeah, yeah, like… what I’m saying is, like,
565 00:51:50.500 ⇒ 00:51:56.430 Awaish Kumar: Creating such kind of, rapport will take me, like, maybe one hour or something.
566 00:51:57.860 ⇒ 00:52:00.710 Awaish Kumar: I just need to keep… for DBT…
567 00:52:00.710 ⇒ 00:52:02.670 Uttam Kumaran: We have the… no, we have the template repo.
568 00:52:04.240 ⇒ 00:52:08.119 Awaish Kumar: I mean, a Python script, basically, for example, which.
569 00:52:08.760 ⇒ 00:52:10.059 Uttam Kumaran: Reports from that repo.
570 00:52:10.750 ⇒ 00:52:21.770 Awaish Kumar: which creates the structure of dbt, which creates the macros we need, the kind of setup we need if it’s a snowflake, we know the standard, like, prod mods, prod intermediate things, we have to…
571 00:52:21.930 ⇒ 00:52:27.490 Awaish Kumar: You need this custom macro file to be there, and so it will automatically have all this set up in there.
572 00:52:27.640 ⇒ 00:52:30.900 Awaish Kumar: And then you just start adding new stuff.
573 00:52:32.000 ⇒ 00:52:40.980 Uttam Kumaran: Yeah, maybe one other way you could do it is, like, for example, for every client, I want to create a new repo and put in the pro… put in the SOWs and the project plans.
574 00:52:41.250 ⇒ 00:52:45.049 Uttam Kumaran: One thing you could do is, you could even write a cursor
575 00:52:46.480 ⇒ 00:52:53.429 Uttam Kumaran: cursor command that says, create the dbt repo, like, pull this, and then modify it
576 00:52:53.990 ⇒ 00:53:04.969 Uttam Kumaran: to adhere to the project plan, like, we… you can tell we mentioned Stripe, we mentioned this, we mentioned this, go to create those folders, create the Mars folders. Maybe it doesn’t have to create… it doesn’t have to, like.
577 00:53:05.380 ⇒ 00:53:07.929 Uttam Kumaran: Create anything, but… yeah.
578 00:53:09.860 ⇒ 00:53:11.060 Awaish Kumar: Okay, yeah.
579 00:53:15.800 ⇒ 00:53:16.640 Awaish Kumar: Bye.
580 00:53:19.320 ⇒ 00:53:21.950 Awaish Kumar: Is there anything else I should be looking at?
581 00:53:23.590 ⇒ 00:53:24.789 Uttam Kumaran: I think that’s it.
582 00:53:30.640 ⇒ 00:53:32.350 Uttam Kumaran: Okay. Cool.
583 00:53:32.850 ⇒ 00:53:33.660 Uttam Kumaran: Alright.
584 00:53:34.190 ⇒ 00:53:35.240 Uttam Kumaran: Thank you, guys.
585 00:53:36.180 ⇒ 00:53:37.280 Uttam Kumaran: I’ll talk to you soon.
586 00:53:38.350 ⇒ 00:53:39.659 Awaish Kumar: Okay, stop. Bye.
587 00:53:39.990 ⇒ 00:53:40.680 Uttam Kumaran: Bye.