Meeting Title: Engineering Leads Sync Date: 2025-08-25 Meeting participants: Shreya Chowdhury, Samuel Roberts, Awaish Kumar, Uttam Kumaran
WEBVTT
1 00:00:07.630 ⇒ 00:00:08.960 Samuel Roberts: There we go.
2 00:00:11.140 ⇒ 00:00:12.619 Shreya Chowdhury: Hey, how’s it going?
3 00:00:12.920 ⇒ 00:00:13.999 Samuel Roberts: Good, how about you?
4 00:00:14.380 ⇒ 00:00:16.230 Shreya Chowdhury: Good, pretty good. ….
5 00:00:16.230 ⇒ 00:00:16.780 Samuel Roberts: Huh?
6 00:00:16.930 ⇒ 00:00:18.589 Shreya Chowdhury: first day, so I’ve just been finishing.
7 00:00:18.590 ⇒ 00:00:19.050 Samuel Roberts: Cheap.
8 00:00:19.270 ⇒ 00:00:27.410 Shreya Chowdhury: stuff, … haven’t officially onboarded or got familiar on the projects that I’ll be working on, but….
9 00:00:27.410 ⇒ 00:00:28.260 Samuel Roberts: Yep.
10 00:00:28.260 ⇒ 00:00:30.560 Shreya Chowdhury: Just logistical stuff so far.
11 00:00:30.560 ⇒ 00:00:35.369 Samuel Roberts: Yeah, no, I’ve been, … It’s been a few weeks now that I’ve been…
12 00:00:36.320 ⇒ 00:00:41.680 Samuel Roberts: Here, so I was kind of jumping into the same, I get it, recently.
13 00:00:42.390 ⇒ 00:00:43.310 Shreya Chowdhury: Yeah.
14 00:00:43.430 ⇒ 00:00:54.510 Shreya Chowdhury: I’m excited, I feel like I’m looking forward to the work. It’s a little bit different than what I was doing on my last job, so I’m excited to see how it goes.
15 00:00:55.210 ⇒ 00:00:55.530 Samuel Roberts: Huh?
16 00:00:55.530 ⇒ 00:00:56.640 Awaish Kumar: Hello, everyone.
17 00:00:56.820 ⇒ 00:00:57.410 Shreya Chowdhury: Hi.
18 00:00:57.410 ⇒ 00:00:57.900 Samuel Roberts: Hello?
19 00:00:59.440 ⇒ 00:01:01.420 Awaish Kumar: How are you doing?
20 00:01:03.300 ⇒ 00:01:04.000 Samuel Roberts: Well….
21 00:01:04.430 ⇒ 00:01:05.230 Shreya Chowdhury: Good.
22 00:01:06.930 ⇒ 00:01:07.710 Awaish Kumar: Okay.
23 00:01:07.960 ⇒ 00:01:16.559 Awaish Kumar: Okay, while Utam joins, like, I’ve shared the document, you can all take a 5 minute to just fill that in, if there’s anything.
24 00:01:19.740 ⇒ 00:01:21.900 Uttam Kumaran: Hello, meeting day! Okay, cool.
25 00:01:21.900 ⇒ 00:01:23.180 Samuel Roberts: Yeah.
26 00:01:23.370 ⇒ 00:01:25.220 Uttam Kumaran: I’m going to that dock.
27 00:01:25.440 ⇒ 00:01:28.630 Shreya Chowdhury: I requested access to it.
28 00:01:32.440 ⇒ 00:01:33.320 Awaish Kumar: Aww.
29 00:01:36.960 ⇒ 00:01:38.590 Uttam Kumaran: Okay, you should be there.
30 00:01:39.710 ⇒ 00:01:41.659 Shreya Chowdhury: Okay, yeah, I’m in now, thank you.
31 00:01:43.230 ⇒ 00:01:45.920 Uttam Kumaran: Okay, give me… yeah, I’ll take just a couple minutes.
32 00:01:45.920 ⇒ 00:01:46.540 Samuel Roberts: Yeah.
33 00:02:52.490 ⇒ 00:02:56.220 Uttam Kumaran: So I’m gonna just… I’ll just… should we just create maybe a section away, like.
34 00:02:56.520 ⇒ 00:02:59.159 Uttam Kumaran: Yeah, maybe we just create 3 sections here.
35 00:03:01.250 ⇒ 00:03:02.050 Uttam Kumaran: -Oh.
36 00:07:45.850 ⇒ 00:07:47.120 Awaish Kumar: Is everyone done?
37 00:07:57.000 ⇒ 00:07:59.800 Uttam Kumaran: I’m… I feel pretty good.
38 00:08:03.300 ⇒ 00:08:04.460 Awaish Kumar: Okay.
39 00:08:05.070 ⇒ 00:08:07.710 Awaish Kumar: Did you want to go first, Clay?
40 00:08:09.240 ⇒ 00:08:12.819 Uttam Kumaran: Yeah, go ahead, if you wanna, lead, then we can start wherever.
41 00:08:14.720 ⇒ 00:08:16.150 Awaish Kumar: Okay, …
42 00:08:19.490 ⇒ 00:08:25.770 Awaish Kumar: Yeah, I… had a… added a comment on A’s, …
43 00:08:26.440 ⇒ 00:08:29.440 Awaish Kumar: So we need AE, like, after…
44 00:08:29.860 ⇒ 00:08:34.099 Awaish Kumar: Kayo, and again, I’m pausing on the Vashtag here.
45 00:08:34.210 ⇒ 00:08:38.000 Awaish Kumar: We do need AE help, I think, and from DE side.
46 00:08:38.360 ⇒ 00:08:40.919 Awaish Kumar: I don’t think we have Zoom.
47 00:08:41.230 ⇒ 00:08:48.740 Awaish Kumar: so much work. Like, we can handle, like, AI team is still… is also doing some DE work, so…
48 00:08:49.660 ⇒ 00:08:52.150 Awaish Kumar: I think on the DE side, we are pretty good.
49 00:08:52.280 ⇒ 00:08:59.210 Awaish Kumar: We need AE help, and then we need… I don’t know, I had a question about Andre, like, if we have any…
50 00:08:59.520 ⇒ 00:09:02.280 Awaish Kumar: … Dashboarding work.
51 00:09:04.280 ⇒ 00:09:10.310 Uttam Kumaran: Yeah, so on the… on the AE side, so we just published a job post, …
52 00:09:10.550 ⇒ 00:09:13.019 Uttam Kumaran: I would love to find, like…
53 00:09:13.280 ⇒ 00:09:16.449 Uttam Kumaran: A junior or mid-level analytics engineer?
54 00:09:16.710 ⇒ 00:09:19.610 Uttam Kumaran: Like, ideally someone that has a…
55 00:09:19.710 ⇒ 00:09:23.959 Uttam Kumaran: we published our thing basically looking for someone with, like, 5 years of DBT experience.
56 00:09:24.380 ⇒ 00:09:30.489 Uttam Kumaran: I mean, 3 is fine, but more interested in people that has worked with the business models that we work with.
57 00:09:30.890 ⇒ 00:09:35.439 Uttam Kumaran: So yeah, I think I’ll have some more answers for you on that probably next week.
58 00:09:35.790 ⇒ 00:09:36.890 Uttam Kumaran: …
59 00:09:37.180 ⇒ 00:09:42.740 Uttam Kumaran: On the DE side, yeah, I mean, I think we should try to bring on that guy you interviewed today.
60 00:09:43.040 ⇒ 00:09:44.429 Uttam Kumaran: or Atomola?
61 00:09:45.600 ⇒ 00:09:46.270 Awaish Kumar: Okay.
62 00:09:46.270 ⇒ 00:09:49.290 Uttam Kumaran: I emailed him, so I’m gonna see whether he’s interested.
63 00:09:50.440 ⇒ 00:09:55.070 Uttam Kumaran: I mean, the DE work is just so easy, but I… I just, like, I just don’t have, like…
64 00:09:55.550 ⇒ 00:09:59.660 Uttam Kumaran: Probably needs at least, like, an hour or two every day, I just don’t have that time, so…
65 00:10:00.270 ⇒ 00:10:04.660 Uttam Kumaran: Hopefully we can find someone. Even, like, a junior person is totally fine.
66 00:10:06.270 ⇒ 00:10:13.639 Uttam Kumaran: On Andre, … Yeah, so I’m sort of waiting to hear back from Robert on…
67 00:10:14.030 ⇒ 00:10:17.410 Uttam Kumaran: Whether we have dashboarding work, …
68 00:10:18.180 ⇒ 00:10:21.909 Uttam Kumaran: I’m also curious to see what Shreya thinks when she starts to see the clients.
69 00:10:22.030 ⇒ 00:10:26.189 Uttam Kumaran: … I think, Shreya, for context, we interviewed, we have, like, a…
70 00:10:26.370 ⇒ 00:10:30.090 Uttam Kumaran: A pretty good offshore,
71 00:10:30.340 ⇒ 00:10:34.030 Uttam Kumaran: dashboard guy, he’s just done a lot of Tableau and Power BI work.
72 00:10:34.170 ⇒ 00:10:35.770 Uttam Kumaran: I know sometimes, like.
73 00:10:35.970 ⇒ 00:10:39.730 Uttam Kumaran: Dashboard work can be really tedious, so we just wanted to try to have someone
74 00:10:39.870 ⇒ 00:10:48.559 Uttam Kumaran: Available for building those. So we can focus on, kind of, client and more, you know, analytics work.
75 00:10:48.660 ⇒ 00:10:49.770 Uttam Kumaran: …
76 00:10:50.000 ⇒ 00:11:06.779 Uttam Kumaran: But I don’t know if we have enough scope immediately available for him, so… I am considering, though, I just need to hear back. I just… I need to hear from the clients that Robert and Trey are gonna be on. None of mine, the ones I’m leading, require it right now.
77 00:11:08.940 ⇒ 00:11:15.060 Shreya Chowdhury: Yeah, okay. I think once I dive in, I can have more of an opinion, on that.
78 00:11:15.530 ⇒ 00:11:16.070 Uttam Kumaran: Okay.
79 00:11:18.940 ⇒ 00:11:24.059 Uttam Kumaran: Cool, and then, yeah, I’ll send a message to… yeah, go ahead, go ahead.
80 00:11:25.500 ⇒ 00:11:29.150 Awaish Kumar: Yeah, I just have, like, you said you have…
81 00:11:29.280 ⇒ 00:11:34.099 Awaish Kumar: A lot of things you have learned which you want to share with the team. Are you considering some demo?
82 00:11:34.230 ⇒ 00:11:36.870 Awaish Kumar: In AI, like, session or something?
83 00:11:38.090 ⇒ 00:11:48.719 Uttam Kumaran: Yeah, I mean, we… I just… I guess, like, we can kind of go to, like… it’s sort of to my next point, is like, I don’t know, I’m just trying to think about how do we build more, like, camaraderie across the engineering team.
84 00:11:49.100 ⇒ 00:11:54.919 Uttam Kumaran: I mean, I… I’m having a blast using Cursor and doing dbt work.
85 00:11:55.230 ⇒ 00:11:57.139 Uttam Kumaran: I think it’s really interesting.
86 00:11:57.380 ⇒ 00:12:04.489 Uttam Kumaran: But, like, I don’t know, I’m just trying to think about, like… ideally, in, like, my past companies, I would have just done a brown bag or, like, something, but…
87 00:12:04.820 ⇒ 00:12:07.709 Uttam Kumaran: I feel like some of the ones we’ve tried, like, have felt
88 00:12:08.220 ⇒ 00:12:12.500 Uttam Kumaran: kind of boring, or like, I don’t know, the discussions have been kind of weak, so…
89 00:12:12.900 ⇒ 00:12:16.590 Uttam Kumaran: I don’t know, I’m, like, interested in, like, running something, but I’m just…
90 00:12:16.980 ⇒ 00:12:18.839 Uttam Kumaran: I’m not… I don’t have, like, a…
91 00:12:19.160 ⇒ 00:12:30.060 Uttam Kumaran: an answer for, like, how to make those engaging, or how people can actually, like, learn. I’m happy to show, I mean, how I use Cursor to write dbt stuff and debug stuff, but…
92 00:12:30.700 ⇒ 00:12:37.630 Uttam Kumaran: I don’t know, kind of just overall… I want it to be, like, a real discussion versus just, like, watching somebody code, you know?
93 00:12:39.420 ⇒ 00:12:42.899 Awaish Kumar: Like, we do have this all-engineering Weekly session, right?
94 00:12:44.290 ⇒ 00:12:48.040 Uttam Kumaran: Yeah, the one last week was good, but it was more of, like, a workshop.
95 00:12:48.650 ⇒ 00:12:49.800 Uttam Kumaran: …
96 00:12:50.210 ⇒ 00:12:57.190 Uttam Kumaran: But I don’t know, like, I don’t know, Shreya or Sam, if you guys ever did anything, like, at your past companies that, like, especially remote.
97 00:12:57.450 ⇒ 00:13:02.889 Uttam Kumaran: around… Professional development or engineering development, like.
98 00:13:03.020 ⇒ 00:13:05.769 Uttam Kumaran: That you thought were actually, like, engaging?
99 00:13:06.450 ⇒ 00:13:22.210 Shreya Chowdhury: Yeah, we used to do these, these demos, and it would be… it became one of those things that got, like, too structured, and then became something that just took time away from other work, but it, like.
100 00:13:22.390 ⇒ 00:13:28.950 Shreya Chowdhury: I think, like, the execution could have been a little better, but in theory, it was just every two weeks, …
101 00:13:29.350 ⇒ 00:13:31.140 Shreya Chowdhury: It would, like…
102 00:13:31.410 ⇒ 00:13:38.550 Shreya Chowdhury: I think we had an hour-long meeting, and then, like, 5 to 7 people would present. You get up to, like, I think…
103 00:13:38.550 ⇒ 00:13:44.719 Shreya Chowdhury: It would be between 1 to 5 minutes of, like, a demo, plus time for questions, and it would just be things like.
104 00:13:44.720 ⇒ 00:13:58.859 Shreya Chowdhury: It was just very casual, like, it wasn’t supposed to be, like, oh, make it very formal and create a presentation for this, but just share what you’ve been working on, where you are, you can share blockers, other people can pitch in, any learnings that you have, something like that.
105 00:14:00.820 ⇒ 00:14:01.590 Uttam Kumaran: Okay.
106 00:14:02.070 ⇒ 00:14:05.959 Uttam Kumaran: Yeah, I mean, I think it’s unique, because we kind of have, like, client work, too.
107 00:14:06.230 ⇒ 00:14:15.079 Uttam Kumaran: But I like that, like, I don’t want to make it anything, like, crazy serious. Like, for the last one, I sort of pulled together stuff, like, right before the meeting.
108 00:14:15.340 ⇒ 00:14:16.889 Uttam Kumaran: I do think that, like.
109 00:14:17.020 ⇒ 00:14:27.629 Uttam Kumaran: sometimes meetings at Brainforce tend to just be, like, me talking, so I want to do at least the least amount of talking, if possible, and I kind of would like it to be, like, engagement from.
110 00:14:28.200 ⇒ 00:14:31.139 Shreya Chowdhury: As many engineers as possible, but yeah.
111 00:14:31.700 ⇒ 00:14:50.519 Shreya Chowdhury: Yeah, or we could also, like, in addition to the meeting, I don’t know if you guys already have, like, a channel or a drive or something like that, where you kind of… we can just share, like, casual threads, like, oh, like, this is what I worked on this week, like, does anyone have any insights, or this is what I learned, like, key takeaways, ….
112 00:14:51.230 ⇒ 00:14:53.730 Awaish Kumar: Yeah, we do have channels for that, right?
113 00:14:54.160 ⇒ 00:14:56.600 Awaish Kumar: AI show and tale, and…
114 00:14:56.880 ⇒ 00:15:02.009 Awaish Kumar: Then another one, Quick Creative Engineering Dev Support, where I share, like, one or two.
115 00:15:02.410 ⇒ 00:15:05.020 Awaish Kumar: Post, and yeah, nothing else.
116 00:15:05.740 ⇒ 00:15:07.610 Uttam Kumaran: Yeah, like, if we look at…
117 00:15:08.080 ⇒ 00:15:12.170 Uttam Kumaran: Like, we do have our data team, channel…
118 00:15:12.350 ⇒ 00:15:17.880 Uttam Kumaran: It’s mostly me. We have, … AI show and tell…
119 00:15:17.990 ⇒ 00:15:23.619 Uttam Kumaran: I like… I share… I do a lot of show and tell.
120 00:15:23.800 ⇒ 00:15:28.660 Uttam Kumaran: I don’t know, I feel like overall, also, like, generally, the team’s pretty shy, so, like.
121 00:15:29.240 ⇒ 00:15:35.679 Uttam Kumaran: I love sharing in here. I mean, I see… I, like, try to do… like, for example, I use Cursor to generate PR reviews.
122 00:15:36.370 ⇒ 00:15:40.520 Uttam Kumaran: PR descriptions last week, I was like, this is the nicest thing ever.
123 00:15:41.180 ⇒ 00:15:52.819 Uttam Kumaran: But, like, I don’t know, I’m… I just, like… I mean, I’m gonna keep talking out loud in these channels, but I would like there to be more engagement. I don’t know what’s the best way to kind of induce that.
124 00:15:53.560 ⇒ 00:15:55.720 Samuel Roberts: Yeah, it’s hard.
125 00:15:55.720 ⇒ 00:15:56.630 Awaish Kumar: Honestly.
126 00:15:59.560 ⇒ 00:16:00.379 Samuel Roberts: Go ahead, Elise.
127 00:16:01.390 ⇒ 00:16:07.800 Awaish Kumar: Okay, so at one of the previous companies, we had this, Like, instead of…
128 00:16:07.960 ⇒ 00:16:19.349 Awaish Kumar: Having a session between all the team members, and … due to, like… like, decrease the shyness.
129 00:16:19.500 ⇒ 00:16:23.370 Uttam Kumaran: Yeah, this Slack app, which basically randomly.
130 00:16:23.370 ⇒ 00:16:24.700 Awaish Kumar: Picks two people.
131 00:16:24.880 ⇒ 00:16:26.869 Awaish Kumar: And so they can talk, maybe once a.
132 00:16:26.870 ⇒ 00:16:27.680 Uttam Kumaran: Oh, nice.
133 00:16:28.160 ⇒ 00:16:36.300 Awaish Kumar: For 30 minutes, so they, like, build… build a relationship together. It doesn’t matter from which team they are, and….
134 00:16:36.300 ⇒ 00:16:36.740 Samuel Roberts: Yeah.
135 00:16:36.740 ⇒ 00:16:43.310 Awaish Kumar: That’s how, like, they can… Like, talk to each other and become more free, and then they…
136 00:16:43.720 ⇒ 00:16:49.409 Awaish Kumar: They can… after that, they can, like, obviously be more open to share their things.
137 00:16:52.450 ⇒ 00:16:53.310 Uttam Kumaran: Okay.
138 00:16:53.940 ⇒ 00:16:58.510 Samuel Roberts: Yeah, I mean, I was gonna… that’s great. I definitely have seen that before, where it just, like…
139 00:16:58.650 ⇒ 00:17:09.609 Samuel Roberts: pairs people up, and you have to… I forget what we called it in DFA. There was a whole app for it that was not just a Slack thing, but I’m sure there are Slack ones, like you said. But I was even just thinking, yeah, just more general, like.
140 00:17:10.630 ⇒ 00:17:16.289 Samuel Roberts: icebreaker-y, like, get-to-know-people thing, because I definitely feel that a little bit. Like, I know Casey Mustafo.
141 00:17:16.290 ⇒ 00:17:17.550 Uttam Kumaran: Yeah, and the problem….
142 00:17:17.550 ⇒ 00:17:23.279 Samuel Roberts: really connect with the other people at all, and it’s just, like, I’m not really sure how best to even do that besides something like a little…
143 00:17:23.470 ⇒ 00:17:28.189 Samuel Roberts: More, like, forced like that, which might be good or might be bad, depending on the person.
144 00:17:29.110 ⇒ 00:17:43.519 Uttam Kumaran: Yeah, probably my biggest feedback is that, like, I… like, I would, like… you guys don’t… don’t even know, like, I would love to… I would… I would just meet with the engineering team every day if I could. I just literally am so low on time that it makes it very hard for me to, like.
145 00:17:43.740 ⇒ 00:17:58.030 Uttam Kumaran: run these, so it has to be something that, like, either this crew can run on some rotation, or something that’s always working. I actually really like that idea of, like, pairing people. I think it’s kind of hard to…
146 00:17:58.340 ⇒ 00:18:03.439 Uttam Kumaran: It’s a lot less intimidating when it’s just, like, you and one other person.
147 00:18:03.680 ⇒ 00:18:07.069 Uttam Kumaran: I also… maybe breakout rooms could work, like.
148 00:18:07.300 ⇒ 00:18:12.390 Uttam Kumaran: We could do the hour-long thing, and then maybe break people out, and then they come back to present something.
149 00:18:12.890 ⇒ 00:18:13.750 Uttam Kumaran: ….
150 00:18:13.750 ⇒ 00:18:14.740 Samuel Roberts: Mmm.
151 00:18:15.530 ⇒ 00:18:32.209 Uttam Kumaran: Yeah, like, I’m interested in camaraderie. I’m also interested in just, like, people being able to build… to see other types of client work. Like, we’ve already seen that people can, like, some of the AI folks are able to do some DE work. I feel like some of the data folks would totally be interested in some of the AI work as well.
152 00:18:32.210 ⇒ 00:18:33.210 Samuel Roberts: Yeah.
153 00:18:33.210 ⇒ 00:18:34.070 Uttam Kumaran: So I’m kind of just trying to.
154 00:18:34.070 ⇒ 00:18:35.000 Awaish Kumar: V….
155 00:18:35.250 ⇒ 00:18:35.610 Uttam Kumaran: Yeah.
156 00:18:35.610 ⇒ 00:18:42.199 Awaish Kumar: So we already have this all-engineering session. On the side, we can also have, This is Andrew.
157 00:18:42.200 ⇒ 00:18:46.999 Uttam Kumaran: Like a once a week, 30 minutes with somebody. Yeah. Like, kind of like a coffee thing.
158 00:18:47.220 ⇒ 00:18:49.620 Samuel Roberts: Yeah, yeah, I think that’s smart.
159 00:18:50.110 ⇒ 00:18:54.130 Awaish Kumar: And after that, we can see how the engagement goes in that all-engineering session.
160 00:18:55.660 ⇒ 00:18:56.120 Uttam Kumaran: Okay.
161 00:18:56.120 ⇒ 00:19:03.289 Samuel Roberts: The other thing to consider, and, you know, I wanna say this delicately, but, like, you, Utam, as someone who is excited, and the CEO, and
162 00:19:03.820 ⇒ 00:19:07.849 Samuel Roberts: talkative, might not, like, I’m not saying you’re making people shy, but I can definitely.
163 00:19:07.850 ⇒ 00:19:17.230 Uttam Kumaran: No, I totally am. Oh my gosh, no, I know, it’s so… I don’t want to, like, I don’t want to be… I would rather not say anything, or I can not even be there.
164 00:19:17.230 ⇒ 00:19:17.920 Samuel Roberts: I’m saying, like.
165 00:19:17.920 ⇒ 00:19:19.080 Uttam Kumaran: I’m more interested, yeah.
166 00:19:19.080 ⇒ 00:19:30.139 Samuel Roberts: That might be… that might be something to try, like, if you… if you… I mean, you are that busy, like, it’s not a bad thing to take, like, something off of your calendar, potentially, that, like, maybe it would be a little bit less, ….
167 00:19:30.140 ⇒ 00:19:30.640 Uttam Kumaran: Okay.
168 00:19:30.640 ⇒ 00:19:36.920 Samuel Roberts: formal if… I mean, I don’t want you to not necessarily come to, like, the all-engineering thing, but, like, if that would maybe get people feeling a little.
169 00:19:36.920 ⇒ 00:19:47.249 Uttam Kumaran: You guys have no idea. I would love to go to that meeting and just listen. Like, I’m not interested in talking more than I need to. But also, if my presence there is in itself, like.
170 00:19:47.620 ⇒ 00:19:52.129 Uttam Kumaran: then I should just… I’ll just not be there. I mean, I’ll be sad, but I won’t… I shouldn’t be there then.
171 00:19:52.130 ⇒ 00:20:02.279 Samuel Roberts: Yeah, I think it’s definitely, like, worth a shot, just to see. Like, I mean, I don’t want to speak for, like, the other people on the team kind of thing, but, like, I definitely can see how that could be a, like, you know, it’s a big meeting….
172 00:20:02.280 ⇒ 00:20:08.250 Uttam Kumaran: I’m kind of a fill-the-air type of person in some of these calls, because otherwise, like, it’s gonna be so awkward, and….
173 00:20:08.250 ⇒ 00:20:09.000 Samuel Roberts: Yeah.
174 00:20:09.000 ⇒ 00:20:10.090 Uttam Kumaran: I just sort of…
175 00:20:10.380 ⇒ 00:20:16.009 Uttam Kumaran: try to, like, fill the air, but… we could try that for this Thursday, like, if… if…
176 00:20:16.140 ⇒ 00:20:26.069 Uttam Kumaran: if y’all want to run Thursday, and I can either help you… I can help set up this, like, one-on-one coffee chat, or if you want to run Thursday.
177 00:20:26.490 ⇒ 00:20:36.900 Uttam Kumaran: breakout rooms, and I can totally just bounce, not be there, and, like, we can catch up on Monday and see how… see how it is, or… yeah, like, that’s fine.
178 00:20:37.130 ⇒ 00:20:43.290 Samuel Roberts: I think both are worth a try, because I think the combination of, like, a time when, like, all engineering meets, but also, like.
179 00:20:43.990 ⇒ 00:20:58.280 Samuel Roberts: pairing up, doing a coffee thing where, like, I now would know someone else, or I know someone across the other team, or, you know, have chatted with someone, got a little more insight into their, you know, work, or their lives in general, like, we’ll make that Thursday
180 00:20:58.740 ⇒ 00:21:06.370 Samuel Roberts: All-engineering thing are probably a little more, … comfortable for everyone, I guess?
181 00:21:07.550 ⇒ 00:21:08.200 Awaish Kumar: Nope.
182 00:21:10.860 ⇒ 00:21:12.889 Samuel Roberts: Yeah, you’re still sharing your screen, by the way.
183 00:21:13.310 ⇒ 00:21:14.469 Uttam Kumaran: Oh, sorry, I’m just….
184 00:21:16.270 ⇒ 00:21:17.880 Samuel Roberts: You’re good, I promise you now.
185 00:21:17.880 ⇒ 00:21:18.520 Awaish Kumar: Clark.
186 00:21:19.530 ⇒ 00:21:20.180 Samuel Roberts: Hold on.
187 00:21:23.350 ⇒ 00:21:25.020 Uttam Kumaran: Okay, I mean, I’m good either way.
188 00:21:29.370 ⇒ 00:21:42.559 Samuel Roberts: Yeah, I would say, I mean, my plan, I think, this Thursday was to do something, just, like, build a simple little web mapping cursor, just to, like, show people, like, what it… how it… I don’t know, I wasn’t really sure exactly… that’s the other thing, because I don’t have a ton of insight into, like, what the other
189 00:21:42.580 ⇒ 00:21:51.319 Samuel Roberts: what the data folks are needing and doing, and, like, how people would use… like, I don’t feel that news either, so I’m… I’m wondering what is, …
190 00:21:52.500 ⇒ 00:21:57.040 Samuel Roberts: the best way to even run that meeting. But, …
191 00:21:57.410 ⇒ 00:22:03.720 Samuel Roberts: I think, like, that is something that can come with time, and especially, I think, if we do a… yeah, like, a random copy thing, exactly.
192 00:22:04.680 ⇒ 00:22:09.200 Samuel Roberts: Like, it might just have to be a overtime thing that meeting gets more comfortable, …
193 00:22:09.660 ⇒ 00:22:16.029 Samuel Roberts: I’m not sure there’s much more to, like, force. You know, the breakouts might be a good idea, because that actually might be a good way to do, like.
194 00:22:17.170 ⇒ 00:22:21.700 Samuel Roberts: kind of that coffee thing on the spot on Thursday, but… …
195 00:22:22.710 ⇒ 00:22:25.220 Samuel Roberts: Yeah, it’s a good thing to maybe try.
196 00:22:25.620 ⇒ 00:22:26.460 Uttam Kumaran: Okay.
197 00:22:27.280 ⇒ 00:22:33.340 Uttam Kumaran: So then, how about this? Like, maybe we’ll set up this coffee thing? I think Awashia, I’ll just pick one and try to get it set up.
198 00:22:33.950 ⇒ 00:22:37.779 Uttam Kumaran: And then, maybe I’ll let you guys run the Thursday meeting.
199 00:22:39.460 ⇒ 00:22:44.919 Uttam Kumaran: And, yeah, we can discuss in engineering leads, like, how I can be helpful, but then…
200 00:22:45.140 ⇒ 00:22:48.829 Uttam Kumaran: I’m just not gonna plan on attending, then. I’m totally fine with that.
201 00:22:49.050 ⇒ 00:22:53.229 Samuel Roberts: Okay. Yeah, I mean, I… It might go exactly the same, but, like, it’s worth.
202 00:22:53.230 ⇒ 00:22:54.119 Uttam Kumaran: Try it, try it.
203 00:22:54.120 ⇒ 00:22:55.040 Samuel Roberts: Yeah, exactly.
204 00:22:55.040 ⇒ 00:23:00.689 Uttam Kumaran: You don’t know until you try, like, maybe I’m… maybe I’m, like, the adult in the room, and, like, if I’m not there, like.
205 00:23:01.430 ⇒ 00:23:10.079 Uttam Kumaran: things get a lot better, less serious, so that’s fine. I don’t mind. I just… I mean, I’m jealous, like, I want to be in that meeting, because that’s all I want to talk about, but….
206 00:23:10.080 ⇒ 00:23:11.059 Samuel Roberts: No, I get it.
207 00:23:11.060 ⇒ 00:23:13.500 Uttam Kumaran: I’ll watch the recording.
208 00:23:13.500 ⇒ 00:23:18.680 Samuel Roberts: Yeah, no, I just, I mean, I know from experience, especially, like, startup stuff and, like, being the….
209 00:23:19.140 ⇒ 00:23:24.220 Uttam Kumaran: No, I actually am very happy that if people meet without me, it’s like a joy, like, I’m glad
210 00:23:24.560 ⇒ 00:23:26.440 Uttam Kumaran: So, I’m mindful.
211 00:23:26.440 ⇒ 00:23:27.630 Samuel Roberts: I’ll give it a shot.
212 00:23:29.320 ⇒ 00:23:36.890 Uttam Kumaran: Okay, cool. In terms of next things, yeah, so I, …
213 00:23:38.090 ⇒ 00:23:45.820 Uttam Kumaran: I also have been working, on Metaplane a bunch this past week, for data observability and alerting.
214 00:23:45.960 ⇒ 00:23:56.089 Uttam Kumaran: It’s been really helpful on urban stems. Been able to catch things a lot faster, and starting to improve our job times for DBT.
215 00:23:56.330 ⇒ 00:24:00.300 Uttam Kumaran: … I don’t really… I mean, I think…
216 00:24:00.860 ⇒ 00:24:05.359 Uttam Kumaran: Yeah, Eden probably needs some work there. That’s something that could go to…
217 00:24:05.980 ⇒ 00:24:09.810 Uttam Kumaran: Ideally, like, the AE, probably, that’s on that side.
218 00:24:10.180 ⇒ 00:24:20.379 Uttam Kumaran: But, yeah, I guess, like, I don’t know, I probably… we need to think about, as an engineering team, like, alert more about how we’re handling alerting and escalations.
219 00:24:20.510 ⇒ 00:24:22.690 Uttam Kumaran: for issues, like, on the AI side.
220 00:24:22.830 ⇒ 00:24:24.660 Uttam Kumaran: I don’t think we have any…
221 00:24:24.990 ⇒ 00:24:29.939 Uttam Kumaran: Sort of, like, error logging, alerting, pager duty type system.
222 00:24:30.680 ⇒ 00:24:31.410 Uttam Kumaran: You know?
223 00:24:31.700 ⇒ 00:24:41.540 Samuel Roberts: Yeah, no, I was… I was thinking about that when we had the call with Arise the other day, like, they… even though just, like, free tier, had some, like, basic, like, traceability and stuff that just, like…
224 00:24:41.780 ⇒ 00:24:49.360 Samuel Roberts: I don’t think we really have set up, and then we talked a little bit about NADN having it, and I think Casey was going to do something, but I’m sure he got busy with other stuff, so… might need to…
225 00:24:50.250 ⇒ 00:24:53.100 Samuel Roberts: get on top of that a little bit more myself, internet, and see.
226 00:24:53.100 ⇒ 00:24:53.420 Uttam Kumaran: Yeah.
227 00:24:53.420 ⇒ 00:24:56.230 Samuel Roberts: What capabilities are there for that sort of stuff.
228 00:24:56.590 ⇒ 00:24:57.900 Samuel Roberts: Okay.
229 00:25:00.300 ⇒ 00:25:01.320 Samuel Roberts: But yeah, I don’t think there’s anything.
230 00:25:01.320 ⇒ 00:25:09.940 Uttam Kumaran: Trey, did you guys have a lot… did you guys have a lot of, like, alerting before at Shopify for stuff, or did you guys have, like, a choice tool?
231 00:25:10.260 ⇒ 00:25:12.930 Uttam Kumaran: So much alerting, yeah. Okay.
232 00:25:13.500 ⇒ 00:25:20.469 Shreya Chowdhury: … I would say, like, yeah, like, it’s definitely good to have some sort of alerting system, …
233 00:25:20.600 ⇒ 00:25:21.740 Shreya Chowdhury: I think…
234 00:25:22.060 ⇒ 00:25:30.239 Shreya Chowdhury: I don’t know if this is a problem that has a great solution yet. The only thing is it didn’t have discernment, so it would be, like.
235 00:25:30.240 ⇒ 00:25:42.260 Shreya Chowdhury: we would have a lot of, alerts for, like, oh, like, this data model, is flagged, or, like, this one’s broken, this pipeline is broken, whatever. And it would be, like, …
236 00:25:43.350 ⇒ 00:25:50.470 Shreya Chowdhury: everyone would get the alert for every single data model that we had, so it would get excessive. But if we can…
237 00:25:50.490 ⇒ 00:26:07.210 Shreya Chowdhury: like, I don’t know the exact use case here right now, like, from a surface level, but it’s like, if we can have some discernment where, like, yeah, like, in a larger channel that people can mute, the alerts go out, and then specific owners of, you know, the issue or the pipeline or whatever get tagged for…
238 00:26:07.240 ⇒ 00:26:09.120 Shreya Chowdhury: Whichever one they’re owners of.
239 00:26:11.950 ⇒ 00:26:15.600 Uttam Kumaran: Okay. Yeah, I feel like we’re… we just, like, haven’t…
240 00:26:15.850 ⇒ 00:26:24.050 Uttam Kumaran: don’t have, like, someone that’s, like, so opinionated over it. It’s kind of been done piecemeal here and there. I don’t… I really don’t like having, like, alert fatigue.
241 00:26:24.560 ⇒ 00:26:29.790 Uttam Kumaran: Yeah. But it does take, like, iterations to get To get there.
242 00:26:30.620 ⇒ 00:26:31.270 Shreya Chowdhury: Yeah.
243 00:26:31.570 ⇒ 00:26:35.750 Shreya Chowdhury: I also don’t have, like, a really strong opinion on it, it is…
244 00:26:35.910 ⇒ 00:26:43.679 Shreya Chowdhury: I mean, that’s also just, you know, part of work. Like, you take a peek at the alert, if it doesn’t apply to you, it can be ignored.
245 00:26:43.800 ⇒ 00:26:47.940 Shreya Chowdhury: It’s preferable not to have more alerts that you need, but…
246 00:26:48.240 ⇒ 00:26:50.129 Shreya Chowdhury: Again, at the same time, if…
247 00:26:50.460 ⇒ 00:26:55.590 Shreya Chowdhury: I don’t know what’s, like, the lesser of the two evils, to have too many, or, like, none at all there.
248 00:27:00.060 ⇒ 00:27:04.989 Uttam Kumaran: Okay, yeah, it’s… mainly for me, my barometer is, like, did we catch it before a client?
249 00:27:06.390 ⇒ 00:27:06.900 Shreya Chowdhury: Yeah.
250 00:27:06.900 ⇒ 00:27:12.290 Uttam Kumaran: Yeah, so… But, like, setting up alerting is just, like, ….
251 00:27:12.290 ⇒ 00:27:13.620 Shreya Chowdhury: It’s so boring.
252 00:27:13.860 ⇒ 00:27:18.870 Shreya Chowdhury: Yeah, it’s… I think it’s also painstaking. I don’t know the process of setting it up, but I….
253 00:27:19.660 ⇒ 00:27:26.400 Uttam Kumaran: Well, you never get it right, so, like, it’s usually just, like, either way too sensitive or, like, not.
254 00:27:27.060 ⇒ 00:27:32.050 Uttam Kumaran: And… I don’t know, maybe it is just, like, I have to be the one that’s…
255 00:27:32.350 ⇒ 00:27:48.760 Uttam Kumaran: most opinionated about it. Because, like, Metaplane is working, and kind of the way I’m thinking about it is, like, I just want to have, like, an observability stack for any engineering work we do. That could be on the AI side or the data side, right? So, like, anytime we go into a client and we produce something.
256 00:27:48.860 ⇒ 00:28:04.799 Uttam Kumaran: what is our process of, like, having some level of observability and alerting on top of it? Because that’s what gets baked into, like, the prices we charge for great engineering workers. Like, I’m like, you’re getting all these things, and we make sure that we have alerting from the get-go.
257 00:28:06.860 ⇒ 00:28:11.870 Uttam Kumaran: But… I think it’s just something as, like, a crew… here, as a crew, I think
258 00:28:12.170 ⇒ 00:28:16.070 Uttam Kumaran: The four of us can probably decide, like, what we want to see there.
259 00:28:16.680 ⇒ 00:28:21.830 Uttam Kumaran: I mean, I know me and Awash have sort of been dealing with issues on the data side. I think I’m…
260 00:28:22.130 ⇒ 00:28:29.319 Uttam Kumaran: kind of Sam, I think I’ll let you kind of form an opinion of, like, what we need to be mindful of on the AI side, and then maybe, Shreya, you can also think about it.
261 00:28:29.740 ⇒ 00:28:36.709 Uttam Kumaran: it’s sort of like, it becomes a problem, but once it’s a problem, it’s like, usually it’s a client’s, like, mad at us because we didn’t catch something. Yeah.
262 00:28:36.710 ⇒ 00:28:37.530 Samuel Roberts: No, that’s definitely….
263 00:28:37.530 ⇒ 00:28:38.120 Uttam Kumaran: problem.
264 00:28:38.680 ⇒ 00:28:49.530 Samuel Roberts: good to have top of mind now, because I guess I thought of this stuff a little bit more with Arise, and just realizing what we don’t have set up, and I’m like, yeah, okay, this needs to happen. But even just hearing it now is like, okay, we can…
265 00:28:49.770 ⇒ 00:28:54.040 Samuel Roberts: I can form some opinions on that, and figure out the best way, and… I mean…
266 00:28:55.220 ⇒ 00:28:59.579 Samuel Roberts: A little bit of, alert fatigue for a minute is not the worst thing, if you can then
267 00:29:00.040 ⇒ 00:29:03.639 Samuel Roberts: tune in a little bit better afterwards, so I’m almost inclined to go, like.
268 00:29:03.850 ⇒ 00:29:07.629 Samuel Roberts: Overkill on it for now, and then dial it back.
269 00:29:08.500 ⇒ 00:29:09.270 Uttam Kumaran: Okay.
270 00:29:09.270 ⇒ 00:29:11.110 Samuel Roberts: But I need to figure out exactly.
271 00:29:11.340 ⇒ 00:29:13.949 Uttam Kumaran: Yeah, I think maybe… figure out how N8N logs.
272 00:29:13.950 ⇒ 00:29:14.410 Samuel Roberts: Yeah.
273 00:29:14.500 ⇒ 00:29:23.750 Uttam Kumaran: I mean, I would also love to see, like, our token usage. Exactly. There’s a bunch of observability stuff on the AI side that I would love to see eventually, you know?
274 00:29:23.750 ⇒ 00:29:40.800 Samuel Roberts: Yeah, yeah, there was some stuff I was even thinking, like, just in general, I’m trying to… I was digging into some co-pilot stuff, co-pilot kit stuff, and I was like, if we’re using that for anything, their observability there that, like, isn’t necessarily N8N, and I was like, we need something probably a little more drop-on to everything that we can just, like, make the calls in.
275 00:29:40.850 ⇒ 00:29:48.070 Samuel Roberts: Because if anything hasn’t, that’s great, like, that’s really helpful for those workflows, but if we’re going to be building anything else that’s, like.
276 00:29:48.400 ⇒ 00:29:54.769 Samuel Roberts: connecting to N8N additionally, I want to make sure that we’re not missing tokens or something somewhere else.
277 00:29:55.490 ⇒ 00:29:55.850 Uttam Kumaran: Yeah.
278 00:29:55.850 ⇒ 00:30:00.999 Samuel Roberts: But… but I… from what… I mean, even just looking at Arise’s, like, free tier, and just… they’re just… it’s kind of…
279 00:30:01.320 ⇒ 00:30:05.900 Samuel Roberts: I forget how he phrased it, but, you know, there’s lots of tools out there that we can probably plug into.
280 00:30:07.470 ⇒ 00:30:13.609 Samuel Roberts: to just get some very basic starting. Like I said, like you said, we don’t really have much at this point.
281 00:30:16.200 ⇒ 00:30:17.150 Uttam Kumaran: Okay.
282 00:30:18.490 ⇒ 00:30:27.649 Uttam Kumaran: And then otherwise, yeah, I think the only thing I wrote otherwise was, yeah, I mean, just curious about, like.
283 00:30:28.030 ⇒ 00:30:29.820 Uttam Kumaran: Product analytics, now that…
284 00:30:29.960 ⇒ 00:30:34.579 Uttam Kumaran: Shrey, you’re on the team, like, I’m sort of hopeful that you can form an opinionated
285 00:30:34.720 ⇒ 00:30:38.420 Uttam Kumaran: perspective on, like, what great product analytics look like. Like, I think…
286 00:30:38.980 ⇒ 00:30:44.579 Uttam Kumaran: between Awash, you know, I think is really mindful on the AEDE side. I think
287 00:30:45.110 ⇒ 00:31:01.680 Uttam Kumaran: you know, Sam is… of course, has a lot of opinions about front-end, back-end, but AI is so new that I’m sort of… I think you’re starting to get a sense of, like, what good AI engineering looks like in the context of our work, which is mostly on the application side, right? We’re not doing, like, fine-tuning or building or trying.
288 00:31:01.680 ⇒ 00:31:02.290 Samuel Roberts: Oh, yeah, yeah.
289 00:31:02.290 ⇒ 00:31:04.920 Uttam Kumaran: So, it is kind of just, like, product…
290 00:31:05.170 ⇒ 00:31:10.129 Uttam Kumaran: A lot of product building and integration work, but product analytics is something that we’re doing
291 00:31:10.330 ⇒ 00:31:22.020 Uttam Kumaran: across, now, how many pro- like, we’re doing across Eden, and we’re doing across two other, like, so we’re doing this for several clients, and default, so 4 clients, I think, right now.
292 00:31:22.160 ⇒ 00:31:24.890 Uttam Kumaran: So, I’m just kind of, like, curious to see, like, what…
293 00:31:25.230 ⇒ 00:31:29.149 Uttam Kumaran: great work there looks like. I mean, I’ve… as I mentioned, like, I’ve been…
294 00:31:29.540 ⇒ 00:31:32.339 Uttam Kumaran: I’ve been downstream of that type of work.
295 00:31:32.550 ⇒ 00:31:40.100 Uttam Kumaran: Like, I’m usually consuming amplitude or mixed panel data, And, like, In a warehouse, but…
296 00:31:40.470 ⇒ 00:31:44.790 Uttam Kumaran: not really on the instrumentation side, and then definitely I’m usually not the one doing
297 00:31:45.130 ⇒ 00:31:55.609 Uttam Kumaran: like, the measurement of product funnels, and I think there’s probably a range of documents and processes to build on, like, how we make recommendations, how do we test things, so…
298 00:31:55.920 ⇒ 00:32:02.080 Uttam Kumaran: Yeah, it’s sort of wide open. I’m just, like, kind of curious to see, like, what great work looks like in that… in that arena.
299 00:32:02.360 ⇒ 00:32:09.110 Shreya Chowdhury: Yeah, and I really appreciate the creative autonomy there. I think once I…
300 00:32:09.390 ⇒ 00:32:12.289 Shreya Chowdhury: Finish onboarding and get an idea of, like.
301 00:32:12.490 ⇒ 00:32:20.039 Shreya Chowdhury: how… like, what our foundation, our current foundation for product analytics is. I’ll probably be able to have
302 00:32:20.310 ⇒ 00:32:23.030 Shreya Chowdhury: A more educated opinion on, like.
303 00:32:23.150 ⇒ 00:32:37.019 Shreya Chowdhury: oh, what we can do going forward, what would be great new, like, workflows and processes, like, to implement. And yeah, I think it’ll be good to, build some documentation there so that we can recycle it going forward.
304 00:32:38.090 ⇒ 00:32:41.729 Uttam Kumaran: Okay, great. Yeah, and you know, it’s even… even thinking about, like.
305 00:32:42.110 ⇒ 00:32:50.539 Uttam Kumaran: do we need, like, template decks for how we present work, or template Notion docs that need to be filled out anytime we do an analysis, or, like.
306 00:32:51.010 ⇒ 00:32:56.640 Uttam Kumaran: For example, we put together this, … this…
307 00:32:57.200 ⇒ 00:33:02.180 Uttam Kumaran: like, this document now I’ll share with you… …
308 00:33:04.010 ⇒ 00:33:11.989 Uttam Kumaran: this is now a template, like, we have a… we have, like, a template now we use for all our events tracking, which is, like.
309 00:33:12.380 ⇒ 00:33:31.700 Uttam Kumaran: what are the events? Who, like, you know, so we could start to now, anytime we walk into a new client where they have PA work, okay, great, like, we need to set this up, we need to do this. So kind of just, like, what is the expectations for great work there? Because ultimately, then we start to ask the rest of the team to start to align towards the deliverables, and then…
310 00:33:31.930 ⇒ 00:33:42.930 Uttam Kumaran: the deliverables become more, predictable, right? So if we know, hey, setting something like this takes 5 hours, cool, like, the PM team knows that these are the runbooks anytime we walk into a client.
311 00:33:43.200 ⇒ 00:33:47.540 Uttam Kumaran: … And it’s just, like, how do we… this is, I think, a really…
312 00:33:47.830 ⇒ 00:33:50.260 Uttam Kumaran: The product analytics stuff is really, like.
313 00:33:50.500 ⇒ 00:34:02.719 Uttam Kumaran: customer-facing, like, they… they love the insights we find, so how do we sort of get them more, get them faster, make it more digestible? So, yeah, just as you’re thinking about things.
314 00:34:02.980 ⇒ 00:34:04.759 Uttam Kumaran: I would say think about both
315 00:34:05.170 ⇒ 00:34:22.700 Uttam Kumaran: what you’re doing for one client, but then you’ll start to see multiple clients, and okay, what are some low-hanging fruit ways for us to just streamline this? Because for folks like Henry, you know, for Zoran, for Annie, like, I think a lot of them are just looking for direction.
316 00:34:22.920 ⇒ 00:34:37.609 Uttam Kumaran: And to date, I think we’ve just kind of thrown people in the deep end, but I think this crew now can give a lot of the more junior folks direction on what does great work look like in their given vertical,
317 00:34:38.620 ⇒ 00:34:55.430 Uttam Kumaran: And so I’m really excited. I think a lot of them have been asking for that, and our feedback is typically really reactive, just because it’s… we haven’t had, you know, enough people in each segment, but now that we do have opinions, we can start thinking about, okay, like, what does great work look like across the stack, you know, so….
318 00:34:58.250 ⇒ 00:34:59.690 Shreya Chowdhury: Yeah, definitely.
319 00:35:01.460 ⇒ 00:35:02.120 Uttam Kumaran: Cool.
320 00:35:02.380 ⇒ 00:35:07.330 Uttam Kumaran: Great, and then Sam, do you want to talk about AI stuff? You had a couple notes.
321 00:35:07.700 ⇒ 00:35:14.019 Samuel Roberts: Yeah, I mean, I was just… I saw… I want to talk through graphite versus cursor bugbot stuff, …
322 00:35:14.220 ⇒ 00:35:15.610 Samuel Roberts: Cause I… I…
323 00:35:16.370 ⇒ 00:35:23.260 Samuel Roberts: I finally got Graphite connected, and so I was looking at some of that stuff, and it seems good, but the cursor stuff seems to disappear. Have you noticed that in the GitHub?
324 00:35:23.580 ⇒ 00:35:24.340 Uttam Kumaran: Oh, really?
325 00:35:24.390 ⇒ 00:35:30.110 Samuel Roberts: Well, like, I think when something changes, it… it says…
326 00:35:30.910 ⇒ 00:35:35.450 Samuel Roberts: What was it? I’ll need to pull up one of the PRs. …
327 00:35:36.150 ⇒ 00:35:47.019 Samuel Roberts: I think it was, like, this message is now out of date. Yeah, this comment was marked as outdated, so I’m not sure if they’re not working together well, or if that’s just changing code, but it still made me a little bit.
328 00:35:47.020 ⇒ 00:35:48.389 Uttam Kumaran: Wait, can you share that.
329 00:35:48.390 ⇒ 00:35:52.960 Samuel Roberts: Yeah, yeah, so it’s… Look at this pull request.
330 00:35:53.840 ⇒ 00:35:56.100 Uttam Kumaran: Is it because I pushed a commit after it?
331 00:35:56.430 ⇒ 00:36:00.850 Samuel Roberts: I think it’s something related to that, but I’m not sure why it doesn’t keep it for, like.
332 00:36:01.040 ⇒ 00:36:02.620 Samuel Roberts: But I’m not sure if it was…
333 00:36:05.010 ⇒ 00:36:12.300 Samuel Roberts: Okay, maybe this is… maybe… okay, maybe it is the… I thought the comments below it are different, but it looks like it’s just hiding it.
334 00:36:14.170 ⇒ 00:36:15.760 Samuel Roberts: Okay, maybe I just didn’t click in.
335 00:36:15.760 ⇒ 00:36:16.150 Uttam Kumaran: You may have….
336 00:36:16.150 ⇒ 00:36:16.620 Samuel Roberts: But….
337 00:36:16.620 ⇒ 00:36:18.600 Uttam Kumaran: Maybe I marked it, too.
338 00:36:18.600 ⇒ 00:36:27.049 Samuel Roberts: No, I just… I think it was just the way Graphite was doing it versus the way Cursor was doing it. I was a little bit like, oh, Cursor seems less…
339 00:36:27.550 ⇒ 00:36:29.040 Samuel Roberts: I don’t know.
340 00:36:29.200 ⇒ 00:36:34.549 Samuel Roberts: But honestly, never mind. I think I just need to dig in more to both of them. Now that I’m digging into it, I’m like, oh, maybe.
341 00:36:34.550 ⇒ 00:36:42.290 Uttam Kumaran: Yeah, I guess, Shreya, for your context, like, I… I’m just playing around with, like, … AI PR review.
342 00:36:42.670 ⇒ 00:36:48.429 Uttam Kumaran: applications. I tested, like, Kersher’s Bug Bot, I texted Graphite.
343 00:36:48.650 ⇒ 00:36:52.680 Uttam Kumaran: They’re all okay, like, I would say they’re really, like, last mile.
344 00:36:53.170 ⇒ 00:37:01.360 Uttam Kumaran: things, like, if there’s things that are glaringly bad in the PR, it’ll catch. I still think that, like, we should have it, I mean…
345 00:37:01.660 ⇒ 00:37:05.820 Uttam Kumaran: Even if it saves, like, an hour or two a week.
346 00:37:06.020 ⇒ 00:37:09.170 Uttam Kumaran: It helps you, like, catch some things, and…
347 00:37:09.700 ⇒ 00:37:13.259 Uttam Kumaran: Implement the fix that may have gotten in, like, it’s worth it.
348 00:37:13.480 ⇒ 00:37:19.870 Uttam Kumaran: … So, I think I may just choose… Either cursor or graphite.
349 00:37:20.140 ⇒ 00:37:23.760 Uttam Kumaran: there’s a reptile, there’s… Code Rabbit.
350 00:37:24.580 ⇒ 00:37:29.620 Uttam Kumaran: like, I don’t know, not, like, they’re all kind of, like, 20 or 30 bucks a person per month.
351 00:37:29.750 ⇒ 00:37:34.859 Uttam Kumaran: don’t really care, I just wanted to help speed up our… and improve our PR process, so….
352 00:37:35.290 ⇒ 00:37:36.050 Shreya Chowdhury: Yeah.
353 00:37:36.160 ⇒ 00:37:41.430 Shreya Chowdhury: I’m also not too picky. I… I did used to use,
354 00:37:41.690 ⇒ 00:37:51.549 Shreya Chowdhury: cursor to help with PR stuff, like, formatting PRs, like, double-checking, like, a lot of the… like, once all the code and whatever has been tested, …
355 00:37:51.780 ⇒ 00:37:55.280 Shreya Chowdhury: I’m… yeah, I’m also not strongly opinionated there.
356 00:37:55.280 ⇒ 00:37:55.860 Uttam Kumaran: Okay.
357 00:37:55.860 ⇒ 00:38:09.559 Shreya Chowdhury: I think, yeah, it could be useful to have a tool there. I kind of would just unofficially, like, plug in what I had, and then would have it formatted for me and go back and forth a little bit until I got to what I wanted the PR to have.
358 00:38:10.110 ⇒ 00:38:10.750 Uttam Kumaran: Okay.
359 00:38:12.620 ⇒ 00:38:15.859 Uttam Kumaran: Okay, cool, then I’ll probably just pick one, have it, and then…
360 00:38:16.200 ⇒ 00:38:20.090 Uttam Kumaran: I sort of just want to find, like, I think the other piece I mentioned is, like.
361 00:38:20.220 ⇒ 00:38:24.520 Uttam Kumaran: It would be great to start to stand up, like, cursor rules for different types of work.
362 00:38:24.630 ⇒ 00:38:28.389 Uttam Kumaran: like, for example, I’m doing a lot of dbt work, and I think Cursor would just be, like.
363 00:38:28.520 ⇒ 00:38:33.800 Uttam Kumaran: Probably 20% better if I could just give it some structure on, like, how to debug certain things.
364 00:38:33.910 ⇒ 00:38:35.120 Shreya Chowdhury: Yeah. ….
365 00:38:35.690 ⇒ 00:38:42.189 Uttam Kumaran: like, here’s how to run things in dbt, or, like, here’s… here’s, like, what we need to look for. …
366 00:38:43.600 ⇒ 00:38:49.100 Uttam Kumaran: Yeah, I don’t know. That’s probably the extent of, like, where I see it in, like, my workflows.
367 00:38:49.710 ⇒ 00:38:52.360 Uttam Kumaran: … Yeah.
368 00:38:53.980 ⇒ 00:39:05.240 Samuel Roberts: Yeah, I think… you’re right, just… my… my gut is, like, let’s just have one thing on there, so pick one of them, I guess, and run with it and see how it is. For cursor rules, the, …
369 00:39:07.060 ⇒ 00:39:14.259 Samuel Roberts: I don’t know what resources are out there for those things already, or how custom… I don’t know the DBT stuff very well, but, like, I can imagine
370 00:39:14.610 ⇒ 00:39:18.680 Samuel Roberts: putting together some and sharing it with, like, the AI team, for example, but I don’t wanna…
371 00:39:18.790 ⇒ 00:39:26.920 Samuel Roberts: I don’t… someone’s gotta do it, or we gotta find a resource for it, you know what I mean? Or I guess Cursor could write its own rules, but I’m not sure, you know.
372 00:39:27.890 ⇒ 00:39:28.940 Samuel Roberts: Where do you want to go there?
373 00:39:28.940 ⇒ 00:39:48.139 Shreya Chowdhury: I also think I… yeah, I don’t know exactly what the requirements are for the DBT work you’ve been doing, but I feel like that would also be good to, like, have a boilerplate template for everyone to have that’s doing, DE work, and then as we continue to work on it, we can always iterate and add more rules.
374 00:39:50.100 ⇒ 00:40:04.449 Uttam Kumaran: Yeah, like, it… you know, it usually fits, like, one or two, but it’s like, okay, dbt said that there’s a… there’s some database error, I have to copy the dbt logs, I just paste it in there. I generally kind of guess that it’s, like, one or two things.
375 00:40:04.590 ⇒ 00:40:16.889 Uttam Kumaran: And then I’m like, figure it out. And it goes and it runs a bunch of checks, it runs some dbt code, it isolates it. But it’s kind of like, I also want to dictate how it… like, I’d rather it plan first.
376 00:40:17.050 ⇒ 00:40:28.530 Uttam Kumaran: maybe ask for feedback before it, like, starts running stuff in a CLI, and then after it identifies the problem, it should, like, say, here’s a problem, here’s what I think the fix is, do you want me to move forward? Like.
377 00:40:28.810 ⇒ 00:40:38.679 Uttam Kumaran: That sort of stuff is, like, I think probably the better way to engineer, versus sometimes it’s just been running for, like, 30 minutes, looping on nothing. It’s kind of a complete waste.
378 00:40:38.830 ⇒ 00:40:44.830 Uttam Kumaran: So, for AU work and for DU work, I can definitely say there’s probably, like.
379 00:40:45.580 ⇒ 00:40:50.020 Uttam Kumaran: Probably 5 key work, like, types of tickets each.
380 00:40:50.210 ⇒ 00:40:51.920 Uttam Kumaran: Maybe we just have, like.
381 00:40:52.250 ⇒ 00:40:56.809 Uttam Kumaran: If it looks like this, then walk through these steps. If it looks like this, walk through these steps.
382 00:40:57.290 ⇒ 00:41:01.459 Uttam Kumaran: Because usually for me, it’s like, add a column, create a new model.
383 00:41:01.900 ⇒ 00:41:05.060 Samuel Roberts: Right. Answer a question about a logic piece.
384 00:41:05.430 ⇒ 00:41:11.149 Uttam Kumaran: Or… or… it’s usually, like, here’s a dbt log, like, something’s broken.
385 00:41:14.940 ⇒ 00:41:15.750 Samuel Roberts: Okay.
386 00:41:15.920 ⇒ 00:41:21.159 Samuel Roberts: Yeah, I mean, if it’s… if it’s fairly… I… yeah, again, I don’t know that word very well, but if it sounds… it sounds fairly, like.
387 00:41:21.410 ⇒ 00:41:22.250 Samuel Roberts: you know.
388 00:41:22.700 ⇒ 00:41:27.260 Samuel Roberts: Easy to guide it in the right way if the rules are there.
389 00:41:27.260 ⇒ 00:41:33.539 Uttam Kumaran: I also want to prevent it from hallucinating, like, yes, the other day, I was like, it added some columns that, like, just didn’t exist, and, like.
390 00:41:33.900 ⇒ 00:41:38.910 Uttam Kumaran: it just assumed, like, it was there. So, I don’t know, we just need to harden it a little bit.
391 00:41:39.390 ⇒ 00:41:51.140 Samuel Roberts: Yeah, I’ve been looking a little bit at, like, MCP servers for, like, just getting things to connect to other things. I’m kind of getting to, like, have that knowledge. I don’t know what exists for that, or if that even makes sense, but I don’t know…
392 00:41:51.530 ⇒ 00:42:01.619 Samuel Roberts: you know, I don’t know enough about this type of work to really talk too much about it, but my experience has been, like, pointing it very specifically to the things I’m looking at.
393 00:42:02.260 ⇒ 00:42:05.130 Samuel Roberts: Where… or using, like, a little bit of, like.
394 00:42:05.750 ⇒ 00:42:13.130 Samuel Roberts: I haven’t done a ton with MCP besides documentation, because I’m a little nervous to, like, let it have access to other things, even inside projects and stuff.
395 00:42:13.390 ⇒ 00:42:14.280 Samuel Roberts: But…
396 00:42:14.590 ⇒ 00:42:26.800 Samuel Roberts: if it could, like, look up table, like, know the table information, or something like that, or without having to be fed it, I don’t know. I don’t want to make too many assumptions about how this work goes, because I’m not as familiar with it at this point, but…
397 00:42:27.490 ⇒ 00:42:33.080 Samuel Roberts: I, I… I have found ways around it as I’m working, you know, front and back-end stuff.
398 00:42:33.510 ⇒ 00:42:33.899 Uttam Kumaran: I can’.
399 00:42:33.900 ⇒ 00:42:37.769 Samuel Roberts: But I don’t know how that works for data stuff as well, you know, because you don’t want it to…
400 00:42:38.320 ⇒ 00:42:40.950 Samuel Roberts: necessarily do anything irrevocable, I suppose.
401 00:42:41.160 ⇒ 00:42:42.059 Samuel Roberts: Yeah, okay.
402 00:42:43.900 ⇒ 00:42:58.059 Uttam Kumaran: Yeah, I think just as everybody here is using Cursor, let’s think about some ways, because if we’re having challenges, then the rest of the team is certainly, like, gonna get it wrong. So, if we can think about, like, hey, every repo we create, or every time we work with a data client, like.
403 00:42:58.190 ⇒ 00:43:03.350 Uttam Kumaran: you should use this cursor rules. We can commit it to the repo, that way it’s there.
404 00:43:03.550 ⇒ 00:43:07.349 Uttam Kumaran: … Yeah, it’s probably just the easiest thing to do.
405 00:43:07.760 ⇒ 00:43:09.629 Samuel Roberts: Yeah, for me, like….
406 00:43:09.630 ⇒ 00:43:17.370 Awaish Kumar: writing a pipeline, like, for example, DE work, or… any kind of… Doing some, …
407 00:43:17.540 ⇒ 00:43:21.010 Awaish Kumar: for example, building an app or something, that’s really…
408 00:43:21.340 ⇒ 00:43:29.680 Awaish Kumar: like, easy AI… for AI to understand, but when it comes to, like, data with… underlying business knowledge.
409 00:43:30.110 ⇒ 00:43:38.600 Awaish Kumar: which, like, sometimes we also lack what does it mean? So it’s… it’s hard to, like, model it correctly using AI.
410 00:43:41.290 ⇒ 00:43:41.910 Uttam Kumaran: Okay.
411 00:43:43.900 ⇒ 00:43:49.349 Awaish Kumar: Yeah, like, for pipelines, it’s really easy, like, pull the APIs documentation.
412 00:43:49.770 ⇒ 00:43:53.739 Awaish Kumar: feed to, like, ChatGBT and ask for a Python script.
413 00:43:55.330 ⇒ 00:44:00.249 Awaish Kumar: It can, like, 90% will be… And download correctly.
414 00:44:00.940 ⇒ 00:44:01.590 Uttam Kumaran: Yeah.
415 00:44:04.230 ⇒ 00:44:06.410 Uttam Kumaran: I think the other thing I’m trying to think about is, like.
416 00:44:06.900 ⇒ 00:44:12.160 Uttam Kumaran: How do we go from PR to, like, cursor, taking the first stab at something faster?
417 00:44:12.330 ⇒ 00:44:15.059 Uttam Kumaran: That’s probably where there’s maybe more alpha.
418 00:44:17.780 ⇒ 00:44:22.500 Uttam Kumaran: like, what do we need to smooth that out? Like, do tickets need to be, like, a lot clearer?
419 00:44:23.180 ⇒ 00:44:27.009 Uttam Kumaran: Can we use linear MCP to, like, get the ticket in?
420 00:44:27.620 ⇒ 00:44:32.890 Uttam Kumaran: … Yeah, I think it’s just something for us to think about. I mean, I think…
421 00:44:33.140 ⇒ 00:44:36.529 Uttam Kumaran: We’re probably sitting on, like, 20 or 30% efficiency there.
422 00:44:37.000 ⇒ 00:44:41.889 Uttam Kumaran: But as we each go through different types of work, I think we can start to think about
423 00:44:42.240 ⇒ 00:44:46.100 Uttam Kumaran: Can we build runbooks for different types of like…
424 00:44:46.250 ⇒ 00:44:52.219 Uttam Kumaran: probably 50% of our work is really low-hanging fruit stuff that, like, I think AI could probably do pretty quickly.
425 00:44:52.330 ⇒ 00:44:54.260 Uttam Kumaran: You know, it just has to get executed.
426 00:44:54.500 ⇒ 00:44:55.160 Samuel Roberts: Right.
427 00:44:56.540 ⇒ 00:44:57.450 Samuel Roberts: Yeah, I think…
428 00:44:57.780 ⇒ 00:45:05.280 Samuel Roberts: Yeah, the tickets need to be good if we’re gonna do that. I haven’t really tried the linear MCP yet too much, because I… there’s also just, like, knowing…
429 00:45:05.390 ⇒ 00:45:13.660 Samuel Roberts: where things are going. Like, it’s not like it’s just, like, one repo with a bunch of tickets. There’s, like, things are kind of in different repos, depending on what’s, you know, the…
430 00:45:13.780 ⇒ 00:45:23.429 Samuel Roberts: the client, or the… even, you know, within that, like, Daxter Pipelines is for, like, a bunch of clients, and, like, I’m just not sure how best to keep it knowing where the right things are, …
431 00:45:24.370 ⇒ 00:45:31.600 Samuel Roberts: But that might just be an organizational, like, figure it out, and then it knows, you know, we know where to put it… push it to, or point it to, I should say.
432 00:45:31.920 ⇒ 00:45:36.410 Samuel Roberts: But I don’t know too much about that yet, so I may actually play the lunar MCG a little bit more.
433 00:45:36.410 ⇒ 00:45:37.100 Uttam Kumaran: Yeah.
434 00:45:40.260 ⇒ 00:45:41.010 Samuel Roberts: Bye.
435 00:45:42.520 ⇒ 00:45:42.890 Uttam Kumaran: Okay.
436 00:45:42.890 ⇒ 00:45:43.440 Samuel Roberts: niche.
437 00:45:44.990 ⇒ 00:45:51.620 Uttam Kumaran: Cool, … Anything else we want to discuss?
438 00:45:58.250 ⇒ 00:45:59.570 Awaish Kumar: Okay.
439 00:45:59.820 ⇒ 00:46:09.639 Uttam Kumaran: Great, so I’ll try to set up the coffee thing, and we can test it out with our group. If it’s good, then yeah, we should just try to roll it out. I think probably the whole company may be interested.
440 00:46:09.660 ⇒ 00:46:19.860 Uttam Kumaran: And then, yeah, I think, let’s talk in Slack, and then we can, probably closer to Thursday, y’all can decide what the plan is for that meeting. But yeah.
441 00:46:19.860 ⇒ 00:46:26.530 Samuel Roberts: Oh, the only other thing was that moving the meeting to 10 on Wednesday conflicts with my…
442 00:46:26.650 ⇒ 00:46:28.539 Samuel Roberts: stand up, I think.
443 00:46:29.890 ⇒ 00:46:31.310 Uttam Kumaran: Mmm, okay, yeah, I guess….
444 00:46:32.850 ⇒ 00:46:38.239 Samuel Roberts: I just didn’t know if it was, like, that was the only time that worked for people, or if we could, like, just even bump it half an hour back to 10.30.
445 00:46:39.300 ⇒ 00:46:49.050 Uttam Kumaran: Yeah, either works for me. I also know that, like, if you’re up, the AI team is up pretty early, too, so, like, we don’t have to do the AI meeting where it is.
446 00:46:49.050 ⇒ 00:46:51.550 Samuel Roberts: Okay, that was gonna be another question, yeah.
447 00:46:51.550 ⇒ 00:46:58.750 Uttam Kumaran: Because I know you’re up really early, and so you can move it anywhere, anytime. I just usually don’t… I’m trying not to take meetings before 9.
448 00:46:58.770 ⇒ 00:47:02.029 Samuel Roberts: Cool. Because, inevitably, somebody puts a meeting at 8.
449 00:47:02.030 ⇒ 00:47:02.740 Uttam Kumaran: Anyways….
450 00:47:02.740 ⇒ 00:47:03.810 Samuel Roberts: Right.
451 00:47:03.810 ⇒ 00:47:14.010 Uttam Kumaran: So, like, but I’m also just, like, such, just try to keep it open. But now that you’re running it, yeah, feel free, you can move it whenever. So if you’re open to moving that, then that would be ideal.
452 00:47:14.010 ⇒ 00:47:19.510 Samuel Roberts: Okay, yeah, I will definitely… that is something I was thinking about anyway, so, I think that’d be cool.
453 00:47:19.510 ⇒ 00:47:20.729 Shreya Chowdhury: Cool, alright.
454 00:47:21.370 ⇒ 00:47:24.349 Shreya Chowdhury: Sorry, is this for the engineering lead sync on Wednesday?
455 00:47:24.350 ⇒ 00:47:25.960 Uttam Kumaran: Yes, on Wednesday.
456 00:47:26.240 ⇒ 00:47:29.849 Shreya Chowdhury: That one’s… so right now it’s scheduled for 7AM my time.
457 00:47:29.850 ⇒ 00:47:38.450 Uttam Kumaran: Oh, yeah, okay, so actually, we’re gonna have to change it. I forgot, I thought, for some reason, yesterday, I thought you were in New York, so, okay, we should…
458 00:47:38.610 ⇒ 00:47:41.870 Uttam Kumaran: Probably move it to after….
459 00:47:42.330 ⇒ 00:47:44.750 Shreya Chowdhury: If we could move it up just a little, that would be….
460 00:47:45.050 ⇒ 00:47:50.440 Uttam Kumaran: Well, let’s just do it when I… I don’t want it to be so early, then. I mean, we could do it in the afternoon on Wednesdays, too.
461 00:47:50.610 ⇒ 00:47:52.720 Shreya Chowdhury: Okay, yeah, that would be great, thank you.
462 00:47:52.720 ⇒ 00:47:56.910 Uttam Kumaran: Okay. Then, Awish, maybe let’s plan on afternoon Wednesday. I’m just trying to.
463 00:47:56.910 ⇒ 00:47:57.229 Awaish Kumar: Thank you.
464 00:47:57.230 ⇒ 00:48:01.159 Uttam Kumaran: as much of my middle of the day opium as possible.
465 00:48:01.280 ⇒ 00:48:06.290 Uttam Kumaran: So I can actually get something done, but, okay, so then we’ll move it later in the day, that’ll be fine.
466 00:48:08.170 ⇒ 00:48:08.940 Awaish Kumar: Okay.
467 00:48:10.100 ⇒ 00:48:10.830 Uttam Kumaran: Okay.
468 00:48:10.970 ⇒ 00:48:13.330 Uttam Kumaran: Alright, thank you all, appreciate it.
469 00:48:13.330 ⇒ 00:48:13.970 Samuel Roberts: Yeah.
470 00:48:13.970 ⇒ 00:48:14.650 Shreya Chowdhury: Bye.
471 00:48:14.650 ⇒ 00:48:15.140 Uttam Kumaran: Okay.
472 00:48:15.140 ⇒ 00:48:15.820 Awaish Kumar: Have a good one.