Meeting Title: Engineering Leads Sync Date: 2025-09-15 Meeting participants: Awaish Kumar, Samuel Roberts
WEBVTT
1 00:01:37.350 ⇒ 00:01:38.250 Samuel Roberts: Blue.
2 00:01:45.280 ⇒ 00:01:46.090 Awaish Kumar: Hello.
3 00:01:47.320 ⇒ 00:01:48.410 Samuel Roberts: Are you doing, Elise?
4 00:01:49.520 ⇒ 00:01:51.679 Awaish Kumar: I’m good, how about you?
5 00:01:52.240 ⇒ 00:01:54.720 Samuel Roberts: Yeah, doing alright. It’s a busy day, you know?
6 00:01:56.270 ⇒ 00:01:58.010 Awaish Kumar: Curl. Full.
7 00:01:59.050 ⇒ 00:02:04.259 Awaish Kumar: Here’s a… knock… You can fill out the…
8 00:02:06.710 ⇒ 00:02:09.679 Awaish Kumar: inputs, I can download it over time.
9 00:02:10.449 ⇒ 00:02:11.009 Samuel Roberts: Yep.
10 00:04:43.020 ⇒ 00:04:44.910 Awaish Kumar: Okay, are you ready?
11 00:04:45.550 ⇒ 00:04:46.460 Samuel Roberts: Yes, sir.
12 00:04:48.290 ⇒ 00:04:50.369 Awaish Kumar: Okay, do you have normal fault?
13 00:04:50.640 ⇒ 00:04:51.990 Awaish Kumar: Do you want to go first?
14 00:04:52.780 ⇒ 00:05:06.930 Samuel Roberts: Sure, so, very similar to what we talked about towards the end of last week. I’ve got K2 working on two of the new client hubs, but, they’re kind of doing it in between client work, so I’m hoping to get those done. I’m probably going to set the goal for Friday as well for that.
15 00:05:07.350 ⇒ 00:05:22.919 Samuel Roberts: I’m gonna get on top of the more about that. Along with that, there is, I need to get some time, carve out some time to figure out how the client hubs are really working, and figure out a better streamlined approach, because this is very manual still, and I don’t like that.
16 00:05:24.610 ⇒ 00:05:26.100 Samuel Roberts: I might set a goal for you.
17 00:05:26.800 ⇒ 00:05:28.390 Samuel Roberts: Thursday event.
18 00:05:30.100 ⇒ 00:05:32.409 Samuel Roberts: Let’s just, like, wrap ourselves in here, it’s simple.
19 00:05:33.530 ⇒ 00:05:42.419 Samuel Roberts: And then the platform, I need to do some bug fixes, but a little more Q&A, and hopefully we’ll be able to, like, soft launch that on a different URL this week.
20 00:05:45.040 ⇒ 00:05:47.330 Samuel Roberts: about all I have for AI team stuff.
21 00:05:53.340 ⇒ 00:05:58.745 Awaish Kumar: Okay, yeah, so… Perhaps…
22 00:06:04.450 ⇒ 00:06:13.630 Awaish Kumar: For me, like, I’m mostly… I haven’t introduced, like, bulk tracking mechanism for aiding where we are…
23 00:06:13.870 ⇒ 00:06:17.650 Awaish Kumar: Maybe trying to track Different kind of data.
24 00:06:17.800 ⇒ 00:06:20.300 Awaish Kumar: Issues which are coming in.
25 00:06:20.530 ⇒ 00:06:21.770 Awaish Kumar: Come on.
26 00:06:21.960 ⇒ 00:06:27.060 Awaish Kumar: And, then I want to translate it in some kind of,
27 00:06:28.150 ⇒ 00:06:35.019 Awaish Kumar: automated test, right? So, such that, like, we can track them before the client sees it.
28 00:06:35.180 ⇒ 00:06:41.330 Awaish Kumar: Secondly, I’m also… now I… next step for that is I want to extend it for the… call the clients.
29 00:06:41.810 ⇒ 00:06:47.420 Awaish Kumar: Secondly, I want to, like, focus on insomnia.
30 00:06:47.540 ⇒ 00:06:55.279 Awaish Kumar: I see, like, I… I… For the few meetings, I haven’t tried the… Standoffs, basically…
31 00:06:55.870 ⇒ 00:07:00.249 Awaish Kumar: But, because there was no escalation, also, I…
32 00:07:00.410 ⇒ 00:07:08.810 Awaish Kumar: But I see that, like, we’ve worked on this automation quite a… Of course, like… Weeks ago, and
33 00:07:09.020 ⇒ 00:07:11.900 Awaish Kumar: Kind of things were settled down, and…
34 00:07:12.920 ⇒ 00:07:16.800 Awaish Kumar: And why we are spending still, like, so much time, I don’t know.
35 00:07:17.390 ⇒ 00:07:19.940 Awaish Kumar: And what were the peninsula, maybe…
36 00:07:20.060 ⇒ 00:07:23.389 Awaish Kumar: Go back there and figure out what’s happening.
37 00:07:24.140 ⇒ 00:07:29.750 Awaish Kumar: Apart from that, there is some, like, internal work which is going on.
38 00:07:30.810 ⇒ 00:07:39.339 Awaish Kumar: in Linear, and GitHub, and regarding, like… Tracking of yours, but yeah.
39 00:07:39.670 ⇒ 00:07:44.789 Awaish Kumar: So that’s, like, kind of ongoing… Yeah, that’s right.
40 00:07:46.180 ⇒ 00:07:46.880 Samuel Roberts: Of course.
41 00:07:47.820 ⇒ 00:07:55.580 Awaish Kumar: Most of it is, like, the updates, nothing… there’s no… We block work, urgency.
42 00:07:56.640 ⇒ 00:08:00.530 Awaish Kumar: There’s… I want to discuss if you have anything.
43 00:08:00.720 ⇒ 00:08:03.960 Samuel Roberts: Which is, like, wheel braking, or…
44 00:08:04.020 ⇒ 00:08:07.240 Awaish Kumar: Something, like, really… Alright.
45 00:08:10.470 ⇒ 00:08:13.960 Awaish Kumar: Like, ideally, which is blocking you to do something.
46 00:08:14.800 ⇒ 00:08:21.410 Samuel Roberts: Yeah, I don’t think anything is right now, it’s really just a matter of time, management, and trying to get around some of the client work, but…
47 00:08:21.600 ⇒ 00:08:27.310 Samuel Roberts: You know, the internal stuff kind of takes a backseat to some of the client work, so that’s really the only, like, blocker, but…
48 00:08:27.570 ⇒ 00:08:34.770 Samuel Roberts: I am hopefully being able to carve out some time today and tomorrow to dig into some of that stuff, so I don’t think there’s anything major that I worry about.
49 00:08:34.940 ⇒ 00:08:35.860 Samuel Roberts: Longs.
50 00:08:39.490 ⇒ 00:08:41.020 Awaish Kumar: Great,
51 00:08:45.420 ⇒ 00:08:48.320 Awaish Kumar: Okay, apart from that, like, the…
52 00:08:49.440 ⇒ 00:08:52.549 Awaish Kumar: How do you think, like, the team is doing?
53 00:08:53.180 ⇒ 00:08:54.130 Awaish Kumar: Thank you.
54 00:08:54.590 ⇒ 00:08:56.370 Awaish Kumar: Casey and Mustafo.
55 00:08:56.970 ⇒ 00:09:05.339 Samuel Roberts: Yeah, they’re, I mean, they’re pretty good, they’re getting their work done. I don’t, you know, there haven’t… there have been a few blocks that we’ve been able to alleviate.
56 00:09:05.520 ⇒ 00:09:09.540 Samuel Roberts: For example, like, the evals for…
57 00:09:09.680 ⇒ 00:09:14.319 Samuel Roberts: interlude was running on Dagster, and it needs to get, so, we were able to…
58 00:09:14.500 ⇒ 00:09:20.610 Samuel Roberts: Mustafa was able to get that over onto Windmill, and now it’s going. So, like, we’ve been able to keep moving through some of these blocks.
59 00:09:20.720 ⇒ 00:09:23.489 Samuel Roberts: Pretty well last week.
60 00:09:27.340 ⇒ 00:09:28.010 Samuel Roberts: Yeah.
61 00:09:28.600 ⇒ 00:09:37.090 Awaish Kumar: Okay, yeah, so… And do you think that there are still, like, really motivated.
62 00:09:38.130 ⇒ 00:09:41.559 Samuel Roberts: Yeah, yeah, they seem to be. You know, I think…
63 00:09:43.190 ⇒ 00:09:47.069 Samuel Roberts: You know, the work’s engaging, we’re doing some interesting things, so that’s keeping them motivated.
64 00:09:47.220 ⇒ 00:09:49.240 Samuel Roberts: I don’t think there’s been too much…
65 00:09:50.440 ⇒ 00:09:57.800 Samuel Roberts: to worry about that way yet. I am planning to have a check-in with them in the next week or two, so I do need to get that on the calendar, but…
66 00:09:59.090 ⇒ 00:10:02.279 Samuel Roberts: Besides that, yeah, I’m feeling pretty good about the team.
67 00:10:05.780 ⇒ 00:10:07.440 Awaish Kumar: Okay, cool.
68 00:10:08.550 ⇒ 00:10:16.219 Awaish Kumar: Yeah, from my side, like, Now, in the data leap, mostly, we are… I need MLRA.
69 00:10:18.470 ⇒ 00:10:22.949 Awaish Kumar: Are they kind of only analytics engineer, or DAD?
70 00:10:24.470 ⇒ 00:10:27.520 Awaish Kumar: Things are going well, clients…
71 00:10:28.180 ⇒ 00:10:31.360 Awaish Kumar: We are working on are also going when
72 00:10:32.000 ⇒ 00:10:40.210 Awaish Kumar: Because we are, like, we are pretty much utilizing like, AI for our… Analytics work.
73 00:10:43.170 ⇒ 00:10:50.199 Awaish Kumar: I’m not sure, like, where, like, whenever we… whenever we get a new request,
74 00:10:50.660 ⇒ 00:10:58.520 Awaish Kumar: kind of like, we both are now maybe utilizing ChatGPT, of course, to write some Azure queries for us, and
75 00:10:58.640 ⇒ 00:11:01.250 Awaish Kumar: Build some initials.
76 00:11:01.380 ⇒ 00:11:05.090 Awaish Kumar: Set… set of queries, and then build it.
77 00:11:05.370 ⇒ 00:11:08.120 Awaish Kumar: We… we have to do some validation checks.
78 00:11:09.310 ⇒ 00:11:12.100 Awaish Kumar: And, after that, it goes…
79 00:11:13.960 ⇒ 00:11:18.779 Awaish Kumar: for PR review, and then gets merged, so… I’m not sure what…
80 00:11:19.520 ⇒ 00:11:23.589 Awaish Kumar: What is… what step can be automated, or…
81 00:11:23.840 ⇒ 00:11:27.250 Awaish Kumar: where AI can even help more on that part.
82 00:11:27.250 ⇒ 00:11:28.250 Samuel Roberts: Yeah.
83 00:11:28.810 ⇒ 00:11:34.719 Awaish Kumar: Like, that’s what… Like, for the development, it’s already being used.
84 00:11:34.990 ⇒ 00:11:40.969 Awaish Kumar: Mostly… The only manual thing is that you copy-paste things from,
85 00:11:41.180 ⇒ 00:11:46.719 Awaish Kumar: Linear take care of somewhere, and question, question, that’s it, right?
86 00:11:48.970 ⇒ 00:11:54.010 Awaish Kumar: Only automation could be that it, like, directly reads from linear.
87 00:11:56.290 ⇒ 00:12:01.390 Samuel Roberts: There is a way to do that, I haven’t actually used it yet, but I believe Linear has an MCP server.
88 00:12:02.950 ⇒ 00:12:05.250 Awaish Kumar: Yeah, like, we could just see…
89 00:12:06.290 ⇒ 00:12:07.959 Samuel Roberts: Look at this ticket, kind of thing.
90 00:12:09.500 ⇒ 00:12:15.890 Awaish Kumar: Yeah, like, look at this ticket, or maybe just… Part of the reason.
91 00:12:16.060 ⇒ 00:12:20.119 Awaish Kumar: And… Yeah, like, if we could say, like.
92 00:12:20.630 ⇒ 00:12:26.300 Awaish Kumar: If a ticket is labeled with, With some, like…
93 00:12:26.550 ⇒ 00:12:37.779 Awaish Kumar: For example, we know that it’s a new model, we have filled in the description, we can just label it in linear, maybe ready for cursor, and Cursor sees that.
94 00:12:38.360 ⇒ 00:12:44.539 Awaish Kumar: And whenever… if a record is labeled as ready for cursor, it just generates.
95 00:12:44.540 ⇒ 00:12:44.920 Samuel Roberts: Yeah.
96 00:12:44.920 ⇒ 00:12:47.340 Awaish Kumar: the SLC can itself and create a PR.
97 00:12:47.580 ⇒ 00:12:49.500 Awaish Kumar: That could be… Yeah.
98 00:12:50.420 ⇒ 00:12:53.470 Samuel Roberts: I think that’s a good idea, because, I think…
99 00:12:53.700 ⇒ 00:13:02.090 Samuel Roberts: if we can trigger it from linear would be ideal like that, not just automatically run everything in. Not have to pull everything in. I don’t know…
100 00:13:02.600 ⇒ 00:13:07.790 Samuel Roberts: I know there are cursor agents that can work in the browser, or you can call from the browser.
101 00:13:09.960 ⇒ 00:13:13.999 Samuel Roberts: I’m wondering if there’s a way to set that up to automatically scan for those kinds of tickets.
102 00:13:14.580 ⇒ 00:13:16.070 Samuel Roberts: That would be something we have to look into.
103 00:13:16.590 ⇒ 00:13:20.140 Awaish Kumar: But, like, Castle can read the linear, right?
104 00:13:20.580 ⇒ 00:13:22.610 Awaish Kumar: through MCP or something like that.
105 00:13:25.300 ⇒ 00:13:28.249 Samuel Roberts: I’m wondering if there’s a way to automate that even more, where we don’t have to even
106 00:13:28.450 ⇒ 00:13:30.050 Samuel Roberts: Trigger it, you know?
107 00:13:35.390 ⇒ 00:13:39.590 Samuel Roberts: I’ll have to think about that a little bit, because I’m not sure if there’s a way to trigger those background agents.
108 00:13:43.300 ⇒ 00:13:47.050 Samuel Roberts: Yeah, here it is. Linear workspaces delegation, it’s a background agent. Okay.
109 00:13:47.970 ⇒ 00:13:50.050 Samuel Roberts: Maybe we gotta connect that and find out.
110 00:13:52.110 ⇒ 00:13:52.940 Samuel Roberts: Okay.
111 00:13:54.060 ⇒ 00:13:55.230 Samuel Roberts: That’s a good thought.
112 00:13:56.130 ⇒ 00:14:00.350 Samuel Roberts: I’m not sure how best to… Automate that.
113 00:14:01.210 ⇒ 00:14:04.449 Awaish Kumar: Yeah, it can just do the first round, right? Like…
114 00:14:04.700 ⇒ 00:14:06.410 Samuel Roberts: I think exa- exactly, yeah.
115 00:14:06.410 ⇒ 00:14:14.609 Awaish Kumar: like, linear… like, AI-generated stuff, for data stuff, like, I, like, I wouldn’t…
116 00:14:16.840 ⇒ 00:14:23.110 Awaish Kumar: like, 100% rely on it. Sometimes, like, for the Python scripts or something, like, you can just…
117 00:14:23.260 ⇒ 00:14:36.449 Awaish Kumar: like, create a sign-up page, for example, like, thousands and, like, even millions of people done that, right? AI knows what a sign-up page looks like, and generates all of the features.
118 00:14:36.590 ⇒ 00:14:39.039 Awaish Kumar: But for the data stuff, like, is it…
119 00:14:39.560 ⇒ 00:14:42.590 Awaish Kumar: Things change from company to company E.
120 00:14:42.700 ⇒ 00:14:49.269 Awaish Kumar: from people to people, from even teams to team, right? In a company, the…
121 00:14:49.680 ⇒ 00:14:57.730 Awaish Kumar: I mean, context changes, right? The way they want to see things changes, so it’s really hard to standardize that.
122 00:14:58.280 ⇒ 00:15:03.409 Awaish Kumar: And that is the reason we can’t… I, like…
123 00:15:04.310 ⇒ 00:15:09.150 Awaish Kumar: we can’t rely on that, and just AI to build
124 00:15:09.540 ⇒ 00:15:19.840 Awaish Kumar: like, ready to build something, like, which is ready to ship. It just builds, initial SQL query based on the linear
125 00:15:20.200 ⇒ 00:15:23.730 Awaish Kumar: Ticket, which has all the context.
126 00:15:24.190 ⇒ 00:15:25.570 Awaish Kumar: Yeah.
127 00:15:25.800 ⇒ 00:15:32.470 Awaish Kumar: creates a PR so that Next person which goes in just goes in that branch, and…
128 00:15:32.620 ⇒ 00:15:35.890 Awaish Kumar: Starts from there, yeah, I don’t have to copy anything.
129 00:15:36.260 ⇒ 00:15:37.000 Awaish Kumar: It…
130 00:15:37.360 ⇒ 00:15:37.770 Samuel Roberts: Right.
131 00:15:37.770 ⇒ 00:15:55.420 Awaish Kumar: But it will just… what I’m seeing from here is that it will just force us to write good linear tickets, right? Otherwise, I could just… I can do that, the similar stuff, like, copy everything from linear into cursor, generate an SQL, and…
132 00:15:55.920 ⇒ 00:15:59.599 Awaish Kumar: And create a PR. It will just take me, like, maybe 10 to 15 minutes.
133 00:16:00.290 ⇒ 00:16:03.190 Samuel Roberts: Yeah, but if we can save all that, that’d be a lot better.
134 00:16:04.240 ⇒ 00:16:06.310 Samuel Roberts: Yeah, but the thing is.
135 00:16:06.630 ⇒ 00:16:16.199 Awaish Kumar: that 15 minutes, maybe… I don’t know if that matters, like, if I’m already working on a 3-hour ticket, like, that 15-hour saving is…
136 00:16:16.700 ⇒ 00:16:20.410 Awaish Kumar: I don’t know how, because we are… even if we are not writing new models.
137 00:16:20.550 ⇒ 00:16:25.459 Awaish Kumar: all the time, so I don’t know if that saving… how big of that saving is, but…
138 00:16:25.840 ⇒ 00:16:42.100 Awaish Kumar: that forces us to write linear… better linear tickets. Like, we will say, it’s… like, we don’t put the label ready for cursor until it has really, really good description, right? So that way, we are writing better and better linear tickets.
139 00:16:42.410 ⇒ 00:16:52.729 Awaish Kumar: we, like, because we try to, like, we’ll use AI for that, so let’s… let’s, like, force PM and force client to provide,
140 00:16:52.960 ⇒ 00:16:55.609 Awaish Kumar: 100% of the requirements, like.
141 00:16:55.800 ⇒ 00:17:01.629 Awaish Kumar: It won’t be 100 anyway, but, like, we will just force them to do that, and that way.
142 00:17:01.890 ⇒ 00:17:04.940 Awaish Kumar: Like, the engineering workload can be…
143 00:17:05.079 ⇒ 00:17:08.739 Awaish Kumar: Then we can save, basically, because all the…
144 00:17:09.260 ⇒ 00:17:21.510 Awaish Kumar: QA and question answering and front and, like, to and fro from the engineer to the client, including a PM, that will get reduced, and that is actual savvy.
145 00:17:23.150 ⇒ 00:17:24.940 Samuel Roberts: Yeah, I think so, I think you’re right.
146 00:17:26.440 ⇒ 00:17:31.759 Samuel Roberts: I’m not able to connect linear, because it says it’s already connected to Cursor, but I don’t seem to…
147 00:17:32.120 ⇒ 00:17:37.749 Samuel Roberts: have the ability to do anything with it yet, so I might have to ping Uten, because it looks like he’s the one that connected it.
148 00:17:38.600 ⇒ 00:17:42.819 Samuel Roberts: But I will, I will do that, and find out what’s going on.
149 00:17:44.780 ⇒ 00:17:45.340 Awaish Kumar: Okay.
150 00:17:45.340 ⇒ 00:17:49.520 Samuel Roberts: If you can add a tag for that, I would love to get that… try that out, even, you know?
151 00:17:50.530 ⇒ 00:17:51.240 Awaish Kumar: Yeah.
152 00:17:51.410 ⇒ 00:17:52.790 Awaish Kumar: That would be nice.
153 00:17:54.110 ⇒ 00:17:58.649 Samuel Roberts: I’ll see if I can ping him about that, and see what he sees on his cursor side.
154 00:17:59.960 ⇒ 00:18:04.949 Samuel Roberts: And maybe we can add it. I actually don’t know what the settings look like, so maybe I’ll dig into that. It’s a good idea.
155 00:18:05.910 ⇒ 00:18:07.960 Samuel Roberts: Say…
156 00:18:09.910 ⇒ 00:18:10.610 Awaish Kumar: Okay.
157 00:18:11.260 ⇒ 00:18:13.570 Awaish Kumar: I don’t have anything else.
158 00:18:14.800 ⇒ 00:18:15.680 Awaish Kumar: Wonderful.
159 00:18:16.630 ⇒ 00:18:18.320 Samuel Roberts: Okay, great. Have a good one.
160 00:18:19.340 ⇒ 00:18:20.670 Awaish Kumar: Beautiful, nice.
161 00:18:21.060 ⇒ 00:18:21.740 Samuel Roberts: Bye.