Meeting Title: Uttam_Melina Date: 2025-02-17 Meeting participants: Melina Tsai, Uttam Kumaran
WEBVTT
1 00:00:43.500 ⇒ 00:00:44.900 Uttam Kumaran: Hey? How’s it going.
2 00:00:44.910 ⇒ 00:00:47.299 Melina Tsai: Hi! I’m good, it’s good. How are you.
3 00:00:47.300 ⇒ 00:00:48.989 Uttam Kumaran: Good, nice to meet you. Finally.
4 00:00:48.990 ⇒ 00:00:50.180 Melina Tsai: Nice to meet you, too.
5 00:00:50.810 ⇒ 00:00:52.070 Uttam Kumaran: How’s everything?
6 00:00:52.794 ⇒ 00:00:54.830 Melina Tsai: Yeah, it’s been going. Well, how about you.
7 00:00:55.040 ⇒ 00:00:56.910 Uttam Kumaran: Good. How’s the how was the weekend.
8 00:00:58.126 ⇒ 00:01:01.300 Melina Tsai: It was pretty a little busy. But yeah.
9 00:01:01.300 ⇒ 00:01:02.510 Uttam Kumaran: Okay, cool.
10 00:01:02.660 ⇒ 00:01:06.159 Uttam Kumaran: Yeah, I I just, I was in Argentina last week.
11 00:01:07.250 ⇒ 00:01:09.829 Uttam Kumaran: So I just got back on Friday. And then.
12 00:01:10.446 ⇒ 00:01:13.913 Uttam Kumaran: yeah, just really like, relax this weekend. So we have a lot of
13 00:01:14.550 ⇒ 00:01:21.830 Uttam Kumaran: we’re pretty busy at work these days. So I’m spending a lot of my time these days back in like Vs code doing data work which is good.
14 00:01:22.120 ⇒ 00:01:25.040 Uttam Kumaran: I was doing a lot of like business stuff for a while and
15 00:01:25.560 ⇒ 00:01:29.570 Uttam Kumaran: sort of getting bored of that. So I’m glad to be back doing like data stuff.
16 00:01:31.790 ⇒ 00:01:41.870 Uttam Kumaran: But yeah, I know I we tried to get in touch last month, and we just got insanely busy, which is good. But we’re still really, really interested in bringing on.
17 00:01:42.040 ⇒ 00:01:45.680 Uttam Kumaran: you know, more people onto our, you know, analyst team.
18 00:01:46.308 ⇒ 00:02:02.721 Uttam Kumaran: So yeah, I just I guess I was kind of curious to just refresh sort of your conversation with Robert, and also just learn a bit more about you and and chat about sort of what you’re up to now, and you know, could happy to, of course, answer any questions about Brainforge
19 00:02:03.230 ⇒ 00:02:14.770 Uttam Kumaran: But I think you know probably know a little bit from talking. Previously, Robert. But yeah, maybe if you want to just give maybe a brief introduction, and I’m I’m happy to share a bit about myself as well, and we can go from there.
20 00:02:15.686 ⇒ 00:02:24.430 Melina Tsai: Okay. So I am a recent graduate from the Cooper Union. And I did a degree in engineering with the focus of machine learning, computer science.
21 00:02:24.590 ⇒ 00:02:44.399 Melina Tsai: And so, like, my personal interests are kind of in the machine learning and mathematics realm. And as well as I’m I’m kind of in the sustainability space as like in the future. That’s where I would eventually want to apply those skills, my like machine learning and math to sustainability. But that’s like a kind of long term vision thing.
22 00:02:45.089 ⇒ 00:02:57.960 Melina Tsai: And yeah, so like, I’ve done coursework in like deep learning. For instance, machine learning, AI, etc. And yeah, just looking for new fun, challenging problems to solve.
23 00:02:58.310 ⇒ 00:03:04.690 Uttam Kumaran: Cool. So what are you up to like right now? Are you working right now? Are you still? Are you doing analyst stuff like? Tell me your situation.
24 00:03:05.220 ⇒ 00:03:22.689 Melina Tsai: Yeah. So I think, like I’ve been doing, it’s it’s here and there. So I’ve been doing some part time work, like. So I’ve been okay. So I’ve been doing some tutoring to kind of like, keep my skill sharp. So I’ve been tutoring in mathematics so mainly like algebra and discrete math.
25 00:03:22.860 ⇒ 00:03:32.829 Melina Tsai: and then also. In the meantime, also just keeping busy and kind of been going to some hackathons the most recent one was like
26 00:03:32.960 ⇒ 00:03:40.779 Melina Tsai: it was a climate hackathon. And we’re kind of developing an AI agent for trying to identify counterparty risk.
27 00:03:40.890 ⇒ 00:03:52.019 Melina Tsai: So yeah, so that’s kind of been, and we’re like, in in the midst of working with the sponsor company to try to extend the project. But yeah, that’s still ongoing. So yeah.
28 00:03:52.530 ⇒ 00:03:56.850 Uttam Kumaran: Nice. What? So how did you guys get? How’d you end up getting in touch with Robert originally.
29 00:03:57.320 ⇒ 00:04:06.690 Melina Tsai: Yeah, it was mostly, I think we had a mutual contact. And yeah, just got connected and had a chat about Brainforge. And I was excited to learn about it.
30 00:04:06.690 ⇒ 00:04:17.579 Uttam Kumaran: Nice. Yeah, I mean, I think you know, your background is sort of the direction we’re probably gonna add longer term. Anyways, I think we do. So you probably know we do for sort of like, full stack data analytics.
31 00:04:17.942 ⇒ 00:04:29.639 Uttam Kumaran: What does that mean? It’s like we do data engineering where we move data from source systems into like a data warehouse like Snowflake. Typically we do data modeling a lot of SQL, so we write a ton of SQL to basically move
32 00:04:29.700 ⇒ 00:04:51.219 Uttam Kumaran: from raw data to what we call like data marts. Think of it like you sell on like 10 different platforms. And you want to build a consolidated sales table which has all of your orders, your revenue, your customers. There’s a lot of sequel that has been written to sort of bring that all together. And then we do both building dashboards and analysis. And really, the long term, we want to push
33 00:04:51.640 ⇒ 00:05:00.690 Uttam Kumaran: way towards doing tougher and tougher things on the analysis side. Right? We work with brands that are making, you know, 1020,000,000, and up per year
34 00:05:01.267 ⇒ 00:05:05.019 Uttam Kumaran: and they rely on us heavily for making decisions.
35 00:05:06.170 ⇒ 00:05:13.240 Uttam Kumaran: as of now, I think we basically are helping clients sort of just like use the data they have. And just basically just see like
36 00:05:13.350 ⇒ 00:05:15.690 Uttam Kumaran: run counts and see like how much
37 00:05:16.220 ⇒ 00:05:40.819 Uttam Kumaran: money they’re bringing in, how many customers they have. But we actually want to move towards solving more complicated problems for them and actually moving them to like what we would call a more proactive state, where we’re actively looking at their data and giving them things that they should go implement right, being a true extension of their of their data team, not just waiting for the CEO to say, I want to see this sort of data. But instead proposing tests.
38 00:05:40.820 ⇒ 00:05:51.450 Uttam Kumaran: solutions as well as doing more complicated analysis, churn risk Ltv modeling and more advanced statistical analysis on customers and cohorting and retention.
39 00:05:51.911 ⇒ 00:05:53.518 Uttam Kumaran: So definitely, I think,
40 00:05:54.140 ⇒ 00:06:23.210 Uttam Kumaran: if, like, the goal is to sort of stay and become more technical, we definitely have opportunity to do that. I think a lot of our work right now is really just on the you know, a lot of just basically building a lot of core dashboards for clients, and then starting to work on more advanced analysis like an excel or in a bi tool like, what do you? What do you think about those sorts of things, and I guess, like, do you have any? You have experience, sort of working with any of those sort of analytics tools before.
41 00:06:24.574 ⇒ 00:06:29.590 Melina Tsai: I think of the things you mentioned. I probably have the most experience in sequel.
42 00:06:29.590 ⇒ 00:06:29.910 Uttam Kumaran: Okay.
43 00:06:30.191 ⇒ 00:06:51.309 Melina Tsai: I did. I worked at bank of China for a little bit as an intern, and I worked with Sq a lot. So yeah, I, that would be interesting also. And just I think in general, I’m also interested in analytics, because I like to kind of figure out what the meaning of numbers are, and then try to apply insights and try to like. Come up with recommendations. That’s probably what interests me.
44 00:06:51.560 ⇒ 00:07:02.049 Uttam Kumaran: So then tell me about that. Your current workflows for doing analysis like, what are the tools that you’re using right now? And like, I mean, let’s just take sort of one of the problems like I gave, which is
45 00:07:02.430 ⇒ 00:07:15.180 Uttam Kumaran: like, you have a client that’s sort of trying to understand a particular correlation between like customers and products, like, what are the tools that you would use and like? Walk me through like what your process would be to sort of like, understand.
46 00:07:15.360 ⇒ 00:07:18.429 Uttam Kumaran: you know, to go through like one of those analysis exercises.
47 00:07:20.030 ⇒ 00:07:29.050 Melina Tsai: No, that’s pretty interesting. I guess. I I don’t know. I kind of like specific problems. But.
48 00:07:31.980 ⇒ 00:07:57.700 Uttam Kumaran: So let’s say, let’s say you’re just given a data set of like orders and customers. And let’s say, geographic region. And you’re trying to find you’re trying to just do some analysis, to find out things like concentration things like customer concentration, geographic concentration, product, concentration, like, I guess, like, walk me through like what your sort of process would be. Initially, as you take a look at that data set, and you know, sort of try to glean some answers from that.
49 00:07:58.520 ⇒ 00:08:16.080 Melina Tsai: Okay. So I think in that case, like, so 1, st 1st thing they would do is probably examine the data set, and just like, try to look at the features and see what is available, and then I guess another thing that I would like to do after that is just do conducting some kind of like exploratory data analysis on it.
50 00:08:16.100 ⇒ 00:08:37.899 Melina Tsai: and really especially what appeals to me is probably going through getting some visual information on it. So if we’re thinking about geographic region like, I usually like plotting. And so if I if it’s possible to get any kind of map data, then it’d be nice to correspond map data with the like, whatever we’re evaluating. So.
51 00:08:37.909 ⇒ 00:08:44.279 Uttam Kumaran: And are you doing this like? Are you doing this in like a notebook locally like, where are you actually conducting like, what are the tools you’re using in the process.
52 00:08:44.280 ⇒ 00:08:54.970 Melina Tsai: Oh, yeah. So okay, so a lot of work I’ve done has been on Google Collab. It’s just a fast and easy way to get into it. But just so like
53 00:08:55.080 ⇒ 00:08:59.750 Melina Tsai: just any kind of like Jupiter notebook type thing, but also
54 00:09:00.070 ⇒ 00:09:19.719 Melina Tsai: other tool like, oh, I guess I also typically like use more pandas and numpy. So that that’s like, so like I’ve done a project before using that to do kind of the mapping like with. I think it’s called Geo pandas. So like using that to kind of get map features and then getting some visualizations out of that.
55 00:09:21.240 ⇒ 00:09:25.440 Melina Tsai: Yeah, it’s like, let’s let’s start with that. I think.
56 00:09:26.660 ⇒ 00:09:56.339 Uttam Kumaran: Okay, that makes sense. So to take that example, you know, a lot of the work that we do for clients right now is in business intelligence tools like looker, for example, where you’re selecting dimensions, you’re selecting measures from a data set. You’re building aggregations like in sequel, you’re probably familiar with sums, averages. Right? So you’re building those measurement aggregations and then visualizing data. So have you done like you mentioned some of the stuff for the hackathon like, how much work have you done? Sort of in like data visualization? You know, beyond, just like
57 00:09:56.430 ⇒ 00:10:00.850 Uttam Kumaran: putting up a line chart for time series data like, Have you? Have you
58 00:10:01.040 ⇒ 00:10:07.299 Uttam Kumaran: like, yeah, I guess if you could just explain a little bit about sort of any sort of work that you’ve done in like data visualization world.
59 00:10:09.553 ⇒ 00:10:10.919 Melina Tsai: Sure. So
60 00:10:10.990 ⇒ 00:10:25.809 Melina Tsai: I think I do like I do end up doing a lot of kind of map work. So like I’ve done a project and kind of so we had a there was one for I based in the New York City region. So we were doing a project for the Housing rights initiative.
61 00:10:25.870 ⇒ 00:10:42.850 Melina Tsai: and we were trying to kind of. They were essentially like filing lawsuits against illegal renting. What so we were they would have. We were given some clean data to. And then we also were trying to map that to specific addresses in the boroughs of New York City.
62 00:10:43.040 ⇒ 00:11:02.319 Melina Tsai: So I think a lot of that was just getting so doing, mapping and then plotting those points on the different to see segmenting the New York City into the they’re called Ntas. It’s a really weird, antiquated, like way to segment New York City. But like you can roughly think of them as neighborhoods.
63 00:11:02.510 ⇒ 00:11:22.389 Melina Tsai: So we kind of just, we would put different data types against that. So thinking about like demographic features, so like income levels, or whether what ethnicities were there just to see, to make sure that things are impartial, and that nobody’s being discriminated against so having them having, we did some like
64 00:11:22.520 ⇒ 00:11:30.610 Melina Tsai: heat mapping, and or just like having having that color coding to really show like, get a sense of what’s going on. And
65 00:11:30.830 ⇒ 00:11:43.169 Melina Tsai: so using that we were, we were yeah. So having that tool. It was kind of like. We ended up making this into an installation piece, and that was open to the public for a little bit in a couple like a year or 2 ago.
66 00:11:43.280 ⇒ 00:11:59.689 Melina Tsai: And it’s just to kind of show. Okay, it’s like a fun little thing where you can see. Okay, what about my neighborhood like what kinds of feature, what what features do I see, and how how how many, how dense, for example, like, are there specific regions that there are a lot of claims in
67 00:11:59.910 ⇒ 00:12:04.159 Melina Tsai: and or if like. And we also did some kind of
68 00:12:04.300 ⇒ 00:12:32.539 Melina Tsai: numbers ratios to see like, okay, how like, if a specific neighborhood had a lot of claims. Okay, maybe this region has a lot of more work to do in terms of like figuring out if there’s rental discrimination where and some neighborhoods are better. Where then? Because they would also have claims that were eventually figure out, okay, we don’t. This is not something we need to litigate. So it’s kind of like doing so yeah, it’s like, so that’s like one mapping thing I’ve done. And then another one is more, for
69 00:12:32.790 ⇒ 00:12:35.579 Melina Tsai: we were doing like an Ev charging placement.
70 00:12:35.760 ⇒ 00:12:46.690 Melina Tsai: ev charging station, please, sir, and trying to optimize location. So it was for that one we were kind of we were based. We were thinking about it for the city of Austin.
71 00:12:47.160 ⇒ 00:12:50.610 Melina Tsai: Oh, sorry, not also Dallas, City of Dallas, and
72 00:12:51.150 ⇒ 00:13:00.510 Melina Tsai: so that one was, we were kind of collecting features and also taking map data from open street maps. And then, yeah, just like
73 00:13:00.660 ⇒ 00:13:07.560 Melina Tsai: finding out those features and then figuring out, okay, where is there demand for charging stations? And then where like.
74 00:13:07.670 ⇒ 00:13:11.329 Melina Tsai: where can we support? What is the current supply? And then kind of doing a justification.
75 00:13:11.580 ⇒ 00:13:24.900 Uttam Kumaran: So let’s talk about that example. Like, if you walk me through. Okay, you were given probably some table with like population by like zip code, or or something else like, how are you actually conducting that analysis like, what libraries are you using? And
76 00:13:25.180 ⇒ 00:13:30.940 Uttam Kumaran: actually, how? How are you actually optimizing for where to place those chargers like technically.
77 00:13:31.900 ⇒ 00:13:32.599 Melina Tsai: Yeah,
78 00:13:33.580 ⇒ 00:13:49.639 Melina Tsai: So okay, we had to collect the data from from scratch first.st So like it was, there was a little bit of process of going through and thinking about. Okay, what features do we want to see? But like, we ended up doing demographic data, and also just
79 00:13:49.750 ⇒ 00:13:59.879 Melina Tsai: like so economics, data, demographic data and geographic. So like things like where parking lots are, or if there’s like a coffee shop nearby, those kind of features.
80 00:14:00.000 ⇒ 00:14:17.389 Melina Tsai: And so for modeling specifically like, we use second, learn, and just also various other like modeling machine learning models. So this could be like xg, boost or forest, those so models in that section. So it was.
81 00:14:17.650 ⇒ 00:14:18.530 Melina Tsai: So
82 00:14:18.700 ⇒ 00:14:28.849 Melina Tsai: we’d use it was pandas and numpy for kind of just the data cleaning data, normalizing. And then there were some a little bit of
83 00:14:28.990 ⇒ 00:14:31.479 Melina Tsai: of data joining, but also using pandas.
84 00:14:32.452 ⇒ 00:14:34.710 Melina Tsai: And yeah, for
85 00:14:35.200 ⇒ 00:14:40.969 Melina Tsai: for map, for mapping. It was Geo pandas for like the features sorry for mapping features. And then.
86 00:14:41.424 ⇒ 00:14:51.150 Melina Tsai: yeah, so, and then it’s kind of like using a machine learning model to figure out the what the predictions were for each for the different zip codes, and then
87 00:14:51.800 ⇒ 00:14:56.699 Melina Tsai: using those zip codes to kind of yeah, generate locations. But yeah.
88 00:14:57.120 ⇒ 00:15:02.029 Uttam Kumaran: Okay, so there’s a bunch of mix of like demographic and population data. Was there any constraints on like
89 00:15:02.640 ⇒ 00:15:06.580 Uttam Kumaran: like, actually where you could place them, or it was just sort of just like, Give us
90 00:15:06.970 ⇒ 00:15:13.029 Uttam Kumaran: like that, long or like, how did you actually decide like, were there any constraints on where you could actually put the chargers.
91 00:15:14.364 ⇒ 00:15:22.879 Melina Tsai: Not constraints per se. It was just. It was more about finding out. Like if we where like
92 00:15:22.980 ⇒ 00:15:28.329 Melina Tsai: the countries, were more economic, I guess, in the sense of like they want to make sure that
93 00:15:28.470 ⇒ 00:15:34.710 Melina Tsai: the that if you place a charging station there will people actually use it. So it’s it’s to make sure there’s like a
94 00:15:35.100 ⇒ 00:15:39.529 Melina Tsai: make. The predictions accurate, I guess. But yeah, not not nothing like
95 00:15:40.270 ⇒ 00:15:42.559 Melina Tsai: you can’t place it in this neighborhood.
96 00:15:42.560 ⇒ 00:15:43.180 Uttam Kumaran: Okay.
97 00:15:44.020 ⇒ 00:15:51.719 Uttam Kumaran: okay, great, that makes awesome. So for that, it’s like a lot of stuff in pandas. And then you just visualize it in collab, basically. And you just output is like a data set for them.
98 00:15:52.960 ⇒ 00:15:55.910 Melina Tsai: Yeah. Yeah. And geopen is for mapping. But yeah.
99 00:15:56.700 ⇒ 00:16:02.110 Uttam Kumaran: Okay, that makes sense. Yeah, that’s a great project. I think a lot of our work is sort of a mix of
100 00:16:02.550 ⇒ 00:16:11.179 Uttam Kumaran: basically a lot of it is a mix of, Hey, we have really rich marketing or customer data, and we need to answer like 5 to 10 questions like.
101 00:16:11.360 ⇒ 00:16:18.660 Uttam Kumaran: Where are our most? Where are our most successful customers like what are what are different ways? We can bundle products together.
102 00:16:19.260 ⇒ 00:16:24.539 Uttam Kumaran: Those are the sorts of things that we work for a lot of clients? Basically, how do we help improve their sales and marketing efforts?
103 00:16:25.640 ⇒ 00:16:30.130 Uttam Kumaran: I guess it’s in some ways it’s a similar problem where you do have a list of features of either
104 00:16:30.250 ⇒ 00:16:34.319 Uttam Kumaran: either products or geos or
105 00:16:35.175 ⇒ 00:16:40.019 Uttam Kumaran: like, basically like time of year. You know, different things like that. And you’re sort of trying to map.
106 00:16:40.120 ⇒ 00:16:58.490 Uttam Kumaran: Okay, what? What is statistically significant? And how can we run tests like that? People also want to do tests on like pricing and discounts and things like that so definitely, I think there’s a lot of room for that, I would say. The only concern probably I have is just we do a lot of work on like the eye tools, right where we’re building visualizations that are
107 00:16:59.185 ⇒ 00:17:04.589 Uttam Kumaran: either like charts or graphs where it does take a really heavy understanding of
108 00:17:04.890 ⇒ 00:17:19.900 Uttam Kumaran: like, hey, this is an e-commerce business. What? What do the executives sort of want to see? Okay, they want to see sales. They want to see profitability. I think that’s the only thing I would love to hear like, what do you think about the challenge? There? I mean, I feel really confident about your work and sort of
109 00:17:20.069 ⇒ 00:17:25.279 Uttam Kumaran: okay, given like a pretty hard statistical puzzle. Go figure it out. But I would say, that’s
110 00:17:25.460 ⇒ 00:17:30.629 Uttam Kumaran: probably half of the work that we do. The big portion of the work that we do is still a lot of
111 00:17:31.442 ⇒ 00:17:43.949 Uttam Kumaran: building up like the base level reporting. Does that seem interesting still? Does that seem like they may be like too much on the just day to day, reporting side, I guess. Tell me like what you think about that.
112 00:17:45.740 ⇒ 00:17:53.320 Melina Tsai: I guess I’m curious about like, what kind? So for bi tools, like, what kind of visualizations you guys work with a lot. And yeah.
113 00:17:53.570 ⇒ 00:17:55.389 Melina Tsai: I want to ask that question first.st
114 00:17:55.390 ⇒ 00:18:07.069 Uttam Kumaran: Yeah, so you can think of visualizations like bar graphs line graphs, stack bar area charts. You wanna show things like percent of change over time, velocity.
115 00:18:07.407 ⇒ 00:18:13.569 Uttam Kumaran: We’re not doing a lot of work about like on, like, really like derivatives. But the main thing is, people want to see progress to goals.
116 00:18:13.970 ⇒ 00:18:22.970 Uttam Kumaran: and they want to see change over time and comparisons. So a lot of it is whole numbers. A lot of it is line charts. A lot of is bar charts, I guess what I’m saying is.
117 00:18:24.160 ⇒ 00:18:32.749 Uttam Kumaran: it’s probably less. It’s less technical in that. You’re building dashboards. But probably the the more difficult part is really understanding, like the business.
118 00:18:33.201 ⇒ 00:18:41.349 Uttam Kumaran: Because we have 2 clients that are in e-commerce. They don’t measure their businesses the same like some people care about profitability. Some people care about growth.
119 00:18:41.470 ⇒ 00:19:11.460 Uttam Kumaran: and I think those are nuances that aren’t readily apparent in the data that you actually need to just understand who the client is and what their goals are. Right. I think that’s a challenge that we have with some folks on our team, which is, it’s not just data for data sake, like people are actively looking at the thing and then going to make decisions. And so part of building a great visualization or a great dashboard is you want to convey the to the user what decision they need to make. And it is sort of
120 00:19:11.630 ⇒ 00:19:25.189 Uttam Kumaran: beyond, just like the colors and things like that. It’s an artifact that allows them to quickly isolate for the executive to isolate where they should be spending their time and where they should go. Effect right? And I think that’s the thing that you know, I meet a lot of analysts that
121 00:19:25.390 ⇒ 00:19:30.949 Uttam Kumaran: are great at just like, okay, if we have a statistical problem. But frankly, for a lot of our
122 00:19:31.060 ⇒ 00:19:36.019 Uttam Kumaran: companies that we support, it will be us actually coming up with what those problems are.
123 00:19:36.478 ⇒ 00:19:42.439 Uttam Kumaran: So it’s not as clean as hey? Here’s like the cleanest data set and go run towards it’s actually like.
124 00:19:43.230 ⇒ 00:20:03.349 Uttam Kumaran: okay, given this client and given the products they sell. What are some things we could go after and help them with, you know. That’s really the the need that we have is we’re not looking for analysts that are handed over. We hand over sort of everything. It’s actually it will take a lot of you putting yourself in the shoes of the client
125 00:20:03.450 ⇒ 00:20:07.609 Uttam Kumaran: and thinking through okay and doing research like looking through
126 00:20:07.620 ⇒ 00:20:36.170 Uttam Kumaran: what is the best way that the best you know, marketing teams measure business. They have, like media, skew media mix models. They do different sort of statistical analysis, different things on customer retention, cohorting right? But a lot of that. Our clients expect us to bring to the table because we’re the experts here. Right? So those are the sort of characteristics that you know. I’m I’m really more interested in in this role. And so I guess, like, yeah, I’ll kick it back to you. Like, what do you think about those.
127 00:20:36.810 ⇒ 00:20:40.530 Melina Tsai: Yeah, I guess I would want to say that like, Oh.
128 00:20:40.690 ⇒ 00:21:06.139 Melina Tsai: I mean a lot of projects I’ve done have always been kind of like started out in a very vague and confusing space like. So I’m very familiar with needing to define a project scope. And just like. And also I mentioned, like, we started out not having a data set. So we had to go out and find and think. So like, starting with a brainstorming session and finding out, okay, what do we think we need? What do you we think is relevant? And then seeing if the data exists. So
129 00:21:06.200 ⇒ 00:21:18.799 Melina Tsai: yeah, it’s so there’s a little bit of that. And also, I think, in general, I also do identify with being good at like kind of thinking in the mindset of other like perspectives. So like.
130 00:21:19.680 ⇒ 00:21:27.760 Melina Tsai: it’s just this is a, it’s a really long example. But like it was like, basically
131 00:21:27.940 ⇒ 00:21:30.999 Melina Tsai: in high school, I was very good at like
132 00:21:31.250 ⇒ 00:21:37.199 Melina Tsai: doing so in history class. It’s really good at thinking about this is really okay. This is really.
133 00:21:37.200 ⇒ 00:21:38.460 Uttam Kumaran: Sure sure. No go ahead. Go ahead.
134 00:21:38.460 ⇒ 00:22:08.339 Melina Tsai: Yeah, yeah, like, so basically, like thinking from the perspective of each of the 13 different colony colony states, and then thinking about what’s advantageous for them in terms of like Rhode Island, smaller. So then they would, but they still want to have good representation, whereas they might think, whereas a bigger state would go. Wait a minute. We need more like we have more people to we have to account for. And it’s just like, so I was very good at kind of thinking from a like what? What the
135 00:22:08.440 ⇒ 00:22:15.509 Melina Tsai: thinking, from the perspective of the different states, and what their different needs are so like. I think, like I
136 00:22:15.870 ⇒ 00:22:39.530 Melina Tsai: I feel like I mean, I would need to learn a little bit about the whole e-commerce thing, but, like also in general, like for a lot of the different kind of projects I’ve ever worked on. It has been a lot of just picking up tooling as I go. And yeah, that’s just been true as well, for like internships, anything I’ve gone to like, there’s always been something new, a new skill to learn, and I’ve been good at picking that up.
137 00:22:39.840 ⇒ 00:22:42.990 Uttam Kumaran: Like, let’s take an example. Is there anything you bought online recently?
138 00:22:45.270 ⇒ 00:22:45.850 Uttam Kumaran: Anything.
139 00:22:45.850 ⇒ 00:22:46.860 Melina Tsai: Flight tickets.
140 00:22:47.250 ⇒ 00:22:49.289 Uttam Kumaran: Flight tickets. Okay, I guess.
141 00:22:49.680 ⇒ 00:23:01.619 Uttam Kumaran: I guess. Think of something more from like Amazon, like a like a product consumer package, good or something like that, or you could look around your room like if anything you look around, you see, like I bought it. I got this on Amazon last week, or something.
142 00:23:02.508 ⇒ 00:23:11.110 Melina Tsai: I think I got a I got a like cow print blanket for my friend first, st like a gift exchange. But yeah, I bought that from Amazon.
143 00:23:11.110 ⇒ 00:23:21.349 Uttam Kumaran: That’s a great example. Let’s say we’re working with ABC blankets. Right? Let’s say they’re a huge blanket company they sell. Let’s say they primarily sell in the Us. Let’s say they they make
144 00:23:21.490 ⇒ 00:23:23.880 Uttam Kumaran: 10 million dollars a year selling blankets.
145 00:23:24.060 ⇒ 00:23:27.549 Uttam Kumaran: Right here are a couple of the problems that that company is gonna have
146 00:23:28.100 ⇒ 00:23:35.099 Uttam Kumaran: one. They’re just like they sell Cowper. They probably have hundreds of hundreds of skews, right? Skews, meaning just variations.
147 00:23:35.240 ⇒ 00:23:51.939 Uttam Kumaran: The second, each skew probably has an associated cost. Cow print may be more expensive than no print versus a custom logo. Right? There may be different things. Second thing, those may take different times to ship to the customer. Right? So there’s there’s some complexity there.
148 00:23:52.180 ⇒ 00:24:06.430 Uttam Kumaran: A 3rd thing, the 4th thing they may ship some of their those their own. They may also rely on Amazon to ship some of those right. It’s a concept of called fulfillment. Right? They may ship some, and then Amazon may ship some on their behalf.
149 00:24:06.540 ⇒ 00:24:08.749 Uttam Kumaran: Right? So you have different fulfillment methods.
150 00:24:09.900 ⇒ 00:24:26.999 Uttam Kumaran: Other thing. They may be running discounts right? So let’s say, there’s like a big college football game. They want to run blanket discounts so they may run discounts on a certain products. How much should they discount? When should they run the discount. How far behind the game should they run the discount?
151 00:24:27.623 ⇒ 00:24:34.829 Uttam Kumaran: Does not. The next thing they they may do a lot of marketing right? They probably run ads on Google, Facebook, Snapchat Tiktok.
152 00:24:35.050 ⇒ 00:24:41.750 Uttam Kumaran: which Ad. Is performing great in terms of not only bringing people to the site, but then having them convert.
153 00:24:42.340 ⇒ 00:24:55.599 Uttam Kumaran: which ad platform actually brings people that buy more than one right like, if they want to look at the lifetime value of a customer. Do people on Facebook tend to buy 3 or 4 or versus Tiktok. Maybe they just buy one, and they leave
154 00:24:55.770 ⇒ 00:25:11.139 Uttam Kumaran: right? So that’s another question. They have maybe they want to get into other. They want to get into pillows right? And let’s say they recently started pillows. And now they’re figuring out, okay, we wanna we want to create like a maybe a buy one. Get one free, or you buy 2 pillows, you get a blanket.
155 00:25:11.370 ⇒ 00:25:22.439 Uttam Kumaran: They’re they want to run a test and figure out what’s the best product? Bundling strategy to do? What’s a test like? How should they architect that test, and what volume numbers would need to be statistically significant
156 00:25:22.550 ⇒ 00:25:23.659 Uttam Kumaran: for them to do that.
157 00:25:24.020 ⇒ 00:25:30.159 Uttam Kumaran: See? This is those are like the questions that we’re answering. And that would say, that’s probably 30% of the questions.
158 00:25:30.661 ⇒ 00:25:54.579 Uttam Kumaran: But you can take what would see on Amazon. Is there a simple? This company is selling me a blanket quickly, you know they get complex. And again, that’s just saying, Amazon, what if they sell, shopify? What if they sell in bed bath and beyond? What if they sell on target? What if they ship overseas. Right? So you get this complexity of not only there’s a data engineering problem which is, can we get all the data? There’s a modeling problem which is
159 00:25:54.830 ⇒ 00:26:18.220 Uttam Kumaran: for my modeling team. Can I make sure that our analyst team has a clean data set across all orders, all discounts, all customers? And then can the analyst team effectively answer the first, st second, and 3rd layer questions for the executives? Not only what is my sales, what’s my highest selling blanket? What are our best selling skews?
160 00:26:18.635 ⇒ 00:26:28.379 Uttam Kumaran: What’s like? What’s our profitability? Basically revenue minus our discounts minus the cost of goods to make the blanket. But then they want to say, Okay, which region should we go into next?
161 00:26:28.950 ⇒ 00:26:34.799 Uttam Kumaran: Right? Which? What? Like? What’s the best promotion we’ve ever ran? And what should we run next.
162 00:26:35.780 ⇒ 00:26:52.209 Uttam Kumaran: Those are like the second layer questions that. And then we get to the 3rd layer question. And there is such a room for optimization in these businesses, you’d be surprised that that blanket company is probably making 1020, 30 million, and they’re probably running the whole thing off a couple of spreadsheets
163 00:26:52.961 ⇒ 00:27:09.820 Uttam Kumaran: for us in the data world we’re like, Oh, my God! You could have done so much! You do so much more analysis. And that’s why they hire us right, because they know there’s so much juice left that they can still squeeze, and that every decision they make should be made with data. So that’s a perfect example of, like what
164 00:27:09.960 ⇒ 00:27:16.849 Uttam Kumaran: we have, several clients that sell online. We have some other clients that sell other products. But great example of the type of work that we do.
165 00:27:19.660 ⇒ 00:27:23.040 Melina Tsai: So in that instance, like what? What kind of
166 00:27:23.300 ⇒ 00:27:32.169 Melina Tsai: you were mentioning? Like juice to sweets like, what kind of what? What additional information do you are you? Is your team typically able to gather from their data set.
167 00:27:32.450 ⇒ 00:27:50.549 Uttam Kumaran: Yeah. So typically they’re they have. No, they may have little to no data support internally. So therefore they’re logging into Amazon and looking at their Amazon dashboard. They’re logging into their Facebook ads and looking at all their ads, but they have no way, maybe, of seeing how many ads, what Ad. Resulted in the highest purchase rate.
168 00:27:50.810 ⇒ 00:27:58.609 Uttam Kumaran: So that’s the work that we do. So we come in and bring their data into one place and combine it, and then give them these aggregated tables that show
169 00:27:58.690 ⇒ 00:28:19.909 Uttam Kumaran: a customer lifetime journey from advertisement to page to purchase right. We can also show them which customers return the most right. So we build these great rich data sets. And then we sort of make sure we either build dashboards that can answer this, the low hanging fruit questions. And then we do more point analysis. When it’s like, okay, we want to answer.
170 00:28:19.930 ⇒ 00:28:33.719 Uttam Kumaran: We want to figure something out. We want to, maybe think about a new product line to launch. We want to go into a new category. They give us sort of a an open, ended problem that we go and solve. So we usually have 2 2 different types of analysis. A lot of the stuff we’re doing for clients are still in one, which is just like.
171 00:28:33.770 ⇒ 00:28:39.929 Uttam Kumaran: do they have views of their data every day? That’s accurate and timely right. Can they see their sales every day?
172 00:28:40.310 ⇒ 00:28:46.640 Uttam Kumaran: That’s not common? That’s really it’s not not that easy to do. You’d be surprised. So that’s the sort of thing that we build for
173 00:28:46.790 ⇒ 00:28:48.689 Uttam Kumaran: in the short term. And then long term.
174 00:28:48.920 ⇒ 00:28:55.940 Uttam Kumaran: we’re building the data, analytics infrastructure so that they can answer any question they need within 2024 to 48 h. Right?
175 00:29:00.650 ⇒ 00:29:05.260 Melina Tsai: Feel like there’s a really big out there. Haven’t really thought about.
176 00:29:05.570 ⇒ 00:29:13.150 Uttam Kumaran: So that’s what I hope to leave you a little bit with. That’s the sort of stuff we’re working with. You know. We do a lot of work with software companies and with e-commerce
177 00:29:13.628 ⇒ 00:29:20.309 Uttam Kumaran: but there’s a whole world. And we didn’t talk about any of the logistics like shipping and and inventory forecasting. And all of that
178 00:29:21.270 ⇒ 00:29:23.419 Uttam Kumaran: customer service related data.
179 00:29:23.894 ⇒ 00:29:32.070 Uttam Kumaran: There’s a whole world of of of analytics. But that’s what clients will really realize to take their problem and break it down like that.
180 00:29:32.200 ⇒ 00:29:37.459 Uttam Kumaran: right? And really say, Wow, okay, there is a lot of opportunity for you. And so
181 00:29:37.890 ⇒ 00:29:39.549 Uttam Kumaran: I think that’s that’s where
182 00:29:40.327 ⇒ 00:29:46.010 Uttam Kumaran: you know, there’s a lot of alpha for us to go figure out, and you know that’s what we get paid to do for sure.
183 00:29:47.530 ⇒ 00:30:07.590 Melina Tsai: I see. And how like is it is it that you guys are taking data that they’re paying different vendors already? And they just need to kind of utilize it? Or is it more like you need to build some kind of tracker to figure out if a customer found an ad on Google, and then went to Amazon. Is it is that like which one does it feel more like, or is it both.
184 00:30:07.590 ⇒ 00:30:24.880 Uttam Kumaran: Yeah. So typically there’s out of the box tools. So for example, when they’re using Amazon, Amazon is a service. But Amazon spits out data, right? Like, for example, like when you when let’s say you’re, you’re a content creator. And you make like 5 Youtube videos, you’re gonna get analytics. How many people watch when they watched.
185 00:30:25.252 ⇒ 00:30:48.169 Uttam Kumaran: How do people clicked? All that data just gets spun out like you could hit their Api and get it. So that’s what we do. We hit their Apis, and we bring that data in but we don’t typically do any of the actual implementation of those systems. They’re already running their business. So we typically work with people who are in business. They’re selling. They’re making money, but they’re not using the information to help them optimize or improve.
186 00:30:48.300 ⇒ 00:30:55.239 Uttam Kumaran: That’s where that’s where we come in. So we’re not actually on the, we don’t implement those software tools. We leverage the data that they spit out.
187 00:30:57.060 ⇒ 00:31:03.000 Melina Tsai: Okay? And so like, do you have an example of a company you guys have worked with, and how like.
188 00:31:03.290 ⇒ 00:31:04.730 Melina Tsai: how you use that data.
189 00:31:05.690 ⇒ 00:31:22.760 Uttam Kumaran: Yeah, like, here’s an example. We worked with the company. They’re called pool parts to go. They are a leading in the e-commerce space for selling pool. Related parts like pumps, covers, ladders for your pool, they sell direct to consumer. They also sell in retail and like wholesale.
190 00:31:23.510 ⇒ 00:31:28.030 Uttam Kumaran: one of the big projects we did for them is they have existing contracts with ups.
191 00:31:28.431 ⇒ 00:31:38.089 Uttam Kumaran: Basically, they pay ups to ship their goods to their customers. They had no idea what was in their ups contracts, and then had no idea how they were being charged.
192 00:31:38.442 ⇒ 00:32:04.399 Uttam Kumaran: What we did is we we went and got the contract from ups. We built a model that basically was able to map. Make sure that we could map any sort of order to a to a price for a package, and we get invoiced for packages. So we know what the actual price we got charged was. But the logic is all rules right? So it’s building sequel rules to make sure that we can forecast what our shipping costs are, gonna be we were able to take that forecast, estimate that, hey? They were about to spend
193 00:32:04.400 ⇒ 00:32:22.049 Uttam Kumaran: like hundreds and hundreds of thousands of dollars on shipping costs. I then went to Ups and Fedex and said, Hey, we need this lowered. Otherwise, we’re gonna switch your competitor. Here’s what we’re gonna do in volume next year. Can you offer any discounts they offered us like almost 80 to 90% discounts.
194 00:32:22.463 ⇒ 00:32:26.409 Uttam Kumaran: And we saved them like a couple of 100 grand just on that project alone.
195 00:32:27.254 ⇒ 00:32:42.039 Uttam Kumaran: Which involves looking at all past shipments, also doing a forecast forward for expected shipment volume based on growth factors based on new markets they were expanding into. And then, finally, it’s not only just handing that to them. It’s saying I’ll call it. I’ll call ups.
196 00:32:43.080 ⇒ 00:32:52.279 Uttam Kumaran: We’re going to be the most knowledgeable data people that could talk about their company shipment, so we might as well be in the conversation, and we we are the ones that handle the the negotiation.
197 00:32:54.470 ⇒ 00:32:55.440 Melina Tsai: That is really cool.
198 00:32:55.670 ⇒ 00:33:07.190 Uttam Kumaran: Yeah, so that’s like a specific example. You know, we have clients that ask us to help them say, Hey, we have 10 products which one should we bundle? We have clients that help us say, like, what like can we ab test different pricing?
199 00:33:07.620 ⇒ 00:33:13.389 Uttam Kumaran: You know, I feel like those are like more simple examples. This is one where it’s like end to end, where
200 00:33:13.870 ⇒ 00:33:18.780 Uttam Kumaran: we sort of got the problem found the answer and then actually executed. And basically.
201 00:33:18.920 ⇒ 00:33:21.409 Uttam Kumaran: we found the huge Roi for them, you know.
202 00:33:22.630 ⇒ 00:33:26.290 Uttam Kumaran: So it’s like, it’s it’s less of, you know, analysis, just for
203 00:33:26.530 ⇒ 00:33:31.240 Uttam Kumaran: novelty. It’s more like they’re trying. They’re gonna make. People are gonna change the way
204 00:33:31.350 ⇒ 00:33:39.640 Uttam Kumaran: they work day to day, based on the data, because there’s no there’s no other thing that you can trust other than your gut instinct. Doesn’t you have to rely on data in these businesses?
205 00:33:40.066 ⇒ 00:34:07.789 Uttam Kumaran: And their valuations on data? They compensate people on data they forecast based on data. And so it’s important that we get it right? But also they have a hard time getting people like us to work there. So they they rely on external people to come in. We’ve learned so much from working with multiple e-commerce brands that we’re the we’re coming there. We’re like, we know exactly how to help you guys. Let me tell you the 10 questions you should be asking, and let me go ahead and answer that, for you, you know. So that’s the stuff kind of stuff that we do.
206 00:34:11.630 ⇒ 00:34:20.830 Melina Tsai: So how is there like an average of like? How long you work with a company? Is it like in your span in months like what? What is kind of the lifetime scope of that.
207 00:34:20.830 ⇒ 00:34:28.450 Uttam Kumaran: Yeah. So I. So this, we’ve only been in business since July 2023. I quit my job in April 2023.
208 00:34:29.840 ⇒ 00:34:32.259 Uttam Kumaran: And we got our 1st client in July.
209 00:34:32.580 ⇒ 00:34:36.109 Uttam Kumaran: and so but most of our clients. We’ve gone in the last 6 months.
210 00:34:36.560 ⇒ 00:34:43.579 Uttam Kumaran: so I’m not really sure yet. I I feel like the average will end up being around a year.
211 00:34:44.313 ⇒ 00:34:51.730 Uttam Kumaran: It takes about 3 to 6 months to really lay a great foundational data infrastructure meaning
212 00:34:52.179 ⇒ 00:35:00.940 Uttam Kumaran: sales, data marketing data, customer service data, financials, shipments, logistics. You have those like amazing data tables that have
213 00:35:01.080 ⇒ 00:35:05.619 Uttam Kumaran: rich dimensionality updated all the time, takes a while to get there.
214 00:35:06.105 ⇒ 00:35:16.110 Uttam Kumaran: Then you could start asking a lot more questions. So typically, most of our clients stay with us at least 6 months. And again we keep answering more and more questions for them. So
215 00:35:16.786 ⇒ 00:35:26.820 Uttam Kumaran: we. But we also help clients hire if they want to hire you know, we’re not super cheap, but we’re also like, not that expensive like I. I want us to raise our prices.
216 00:35:27.920 ⇒ 00:35:31.690 Uttam Kumaran: but also for me, I’m more concerned with
217 00:35:32.208 ⇒ 00:35:34.760 Uttam Kumaran: building an amazing data team of people that
218 00:35:35.190 ⇒ 00:35:37.780 Uttam Kumaran: can take learnings for multiple clients.
219 00:35:37.940 ⇒ 00:35:50.950 Uttam Kumaran: And, in fact, like, I want us to be the best data team available for hire like in the Us. Like, I want to know that if someone, if you have a problem with data where the people to call, no matter what kind of problem they have.
220 00:35:51.220 ⇒ 00:36:00.692 Uttam Kumaran: we work the fastest, we’ll deliver it the highest quality. And of course, like we’re all human beings. So we’re we’re great to work with. And that’s that’s really my my goal.
221 00:36:01.090 ⇒ 00:36:22.520 Uttam Kumaran: I think a lot of like what we do for clients we’ve been doing for a long time like I I did it. I graduated, you know, in 2018, and I’ve been a lot of the work I do right now is I did when I graduated. So not much has changed. Like people still want to see these things about their business. But what has changed is our ability to use the latest statistical models.
222 00:36:22.520 ⇒ 00:36:34.069 Uttam Kumaran: the latest data modeling and data engineering tools. And to really tell the story of data, right? And this is what I’m sure you know a lot about is the storytelling aspect of how do you walk through
223 00:36:34.760 ⇒ 00:36:56.699 Uttam Kumaran: the problem you’re trying to solve. Why, it’s important. And then, ultimately, how does that end client actually see that in a way they understand right? Some people, they we have Cfo that we work with that don’t want dashboards. They just want an excel sheet. We have some executives that are like, I need to see a visual right. And so ultimately we’re judged by the end product. However, there’s a lot of work that happens before them that
224 00:36:56.820 ⇒ 00:36:59.560 Uttam Kumaran: customers will will never see, you know.
225 00:37:01.300 ⇒ 00:37:08.799 Uttam Kumaran: But right now it’s hard. It’s hard. A lot of these customers will never be able to hire data folks on their own, or they’ve tried to hire, and it’s not worked out.
226 00:37:08.910 ⇒ 00:37:12.460 Uttam Kumaran: And so that’s where we come in. We really try to save the day for some of these guys.
227 00:37:16.240 ⇒ 00:37:21.429 Melina Tsai: Yeah, I don’t know. That was thanks for just all of that knowledge. I.
228 00:37:21.430 ⇒ 00:37:21.850 Uttam Kumaran: Yeah.
229 00:37:21.850 ⇒ 00:37:24.750 Melina Tsai: I think it was pretty eye opening, regardless.
230 00:37:24.750 ⇒ 00:37:26.780 Uttam Kumaran: What do you think? After hearing all that like.
231 00:37:27.610 ⇒ 00:37:31.839 Uttam Kumaran: yeah, how do you? How do you think about that? Versus sort of some of the stuff that you’ve been doing.
232 00:37:33.011 ⇒ 00:37:41.309 Melina Tsai: I think it’s a pretty interesting space to work in, and I would be curious to take on that challenge. I I don’t know, like
233 00:37:42.170 ⇒ 00:37:44.289 Melina Tsai: I think the like.
234 00:37:44.890 ⇒ 00:37:57.620 Melina Tsai: I mean, granted, like, I’m coming from a more engineering background. But like getting into the kind of business aspect of it would be really interesting. So I I don’t know it would. I would definitely, at the very least, like, I know, it would be a really good growth experience so
235 00:37:57.780 ⇒ 00:38:00.920 Melina Tsai: excited for any kind of like possibilities.
236 00:38:01.620 ⇒ 00:38:10.160 Uttam Kumaran: Cool. Yeah. And I, I studied computer engineering. So I didn’t have any business background. I did some finance in school. But data is a really unique place where
237 00:38:12.160 ⇒ 00:38:23.669 Uttam Kumaran: like, it’s such a you have to know both really? Really? Well, there’s not many engineering jobs where you have to know the business side. In order to be successful this side, you really really have to know. And, in fact.
238 00:38:23.870 ⇒ 00:38:25.629 Uttam Kumaran: over time I’m
239 00:38:25.900 ⇒ 00:38:44.870 Uttam Kumaran: we will remove the need for like project management, because we want our engineers to themselves, be able to understand the clients needs and know what to work on right. So I don’t want filters between our engineering team and the clients. In fact, like, I want people to feel like we are an extension of their business.
240 00:38:45.363 ⇒ 00:38:52.070 Uttam Kumaran: You know. And I think that’s the unique opportunity. And then the other thing about Brainforge is like, you sort of get to work at like 5 companies at once.
241 00:38:52.970 ⇒ 00:39:04.259 Uttam Kumaran: which is great, like, if you’re interested in in like accelerated learning and basically like being like, okay, I, instead of just working at one place and sort of seeing it methodically go. I want to go somewhere where
242 00:39:04.380 ⇒ 00:39:08.457 Uttam Kumaran: we have 7 active clients. We have AI clients.
243 00:39:09.010 ⇒ 00:39:14.999 Uttam Kumaran: you know, we’re we’re using AI internally for for doing engineering sales marketing work.
244 00:39:16.330 ⇒ 00:39:24.190 Uttam Kumaran: You know, we’re sort of balancing everything. It’s a really really cool place to work, I think, what we what we have in terms of that we lack in structure.
245 00:39:24.340 ⇒ 00:39:35.610 Uttam Kumaran: That’s a tough part, like we are an early company and a lot of it is sort of learning on the job. But nothing is sort of technically challenging. I think the challenge is really understand
246 00:39:35.800 ⇒ 00:39:38.880 Uttam Kumaran: priority and understanding, like, what’s the most impactful for clients.
247 00:39:40.930 ⇒ 00:39:54.569 Melina Tsai: So then, like structure, wise tentative or not like is, do you guys typically set different teams on specific companies? Or do you have, like just everyone working on everything at once, like, How how do you do it?
248 00:39:54.570 ⇒ 00:40:15.059 Uttam Kumaran: Yeah, that’s correct. So we do like sort of like client pods, what we call like engagement pods. Basically, we have, every client has one person that’s like an engagement lead. This is sort of like a technical lead. Someone who understands sort of all the technical aspects, but also understands the client. We have a project manager, someone who sort of just make sure that people are assigned tickets there.
249 00:40:15.390 ⇒ 00:40:32.129 Uttam Kumaran: They know what they need to work on next. And then we have all the engineers. So either data, engineers, analytics, engineers or analysts. And then, yeah, we have pods of between 3 to 5 per client. And it’s people work on multiple clients. Right? And so we’re certainly figure right now, we’re we’re sort of around
250 00:40:32.260 ⇒ 00:40:35.789 Uttam Kumaran: 3 on the analytics engineering side. We have
251 00:40:36.030 ⇒ 00:40:39.259 Uttam Kumaran: 3 on the analyst side. 2 of them are part time.
252 00:40:39.689 ⇒ 00:40:43.200 Uttam Kumaran: I sort of float everywhere, and Robert sort of floats everywhere.
253 00:40:44.360 ⇒ 00:40:47.150 Uttam Kumaran: But we’re definitely trying to bring on one more analyst.
254 00:40:47.713 ⇒ 00:41:03.339 Uttam Kumaran: And most likely we’ll bring on 2 there and and one more analytics engineer as well. And ideally again, like, we internally know all the answers for how to solve these data problems. I think our challenge is really context switching.
255 00:41:03.640 ⇒ 00:41:07.639 Uttam Kumaran: And second, is, like being able to
256 00:41:07.780 ⇒ 00:41:19.489 Uttam Kumaran: to prescribe the solutions to the clients. It’s rare that they come to us and tell us, go run the statistical analysis, because they just don’t know that that’s even possible.
257 00:41:20.126 ⇒ 00:41:34.150 Uttam Kumaran: And so for us, it’s up to. It’s up to the technical engagement manager and the team to sit and think. Okay, what do these guys need right now, you know. And so we’re moving as much as we can from sort of reactive to more proactive work where
258 00:41:34.290 ⇒ 00:41:49.409 Uttam Kumaran: we’re saying, Okay, we want to go solve this next problem for you. And sort of that’s the stage that we are. Now. Of course, we’re bringing on more clients as well. So we just need, you know, great people. But last thing I’ll say is like, we’re not trying to hire people in sort of a way where it’s like
259 00:41:50.040 ⇒ 00:41:54.209 Uttam Kumaran: work for a year. And bounce like we really wanted this to be a place where
260 00:41:54.790 ⇒ 00:42:15.750 Uttam Kumaran: the next data challenge is just like one more client away, basically. And you can either take on things harder in terms of leadership and managing people. You can also go deeper technically on one area or another. Also, you could move quickly between a stack like you could go more becoming more on the data engineering side analytics engineering side. There’s a lot of openings.
261 00:42:15.850 ⇒ 00:42:27.670 Uttam Kumaran: Then the luckily the lucky thing about this company is, it’s I’m like an engineer. So there’s like not much bureaucracy, and there’s not much like red tape on anything. In fact, like, I’m trying to push more people to
262 00:42:27.780 ⇒ 00:42:32.009 Uttam Kumaran: think outside the box and take on more responsibilities.
263 00:42:32.528 ⇒ 00:42:39.410 Uttam Kumaran: So that’s that’s most likely how we’ll be for the future. Anyways. So yeah.
264 00:42:46.040 ⇒ 00:42:47.899 Melina Tsai: No, just thinking about the whole like
265 00:42:48.450 ⇒ 00:42:52.410 Melina Tsai: going or getting to like more management versus, like
266 00:42:52.670 ⇒ 00:42:55.969 Melina Tsai: being more domain specific. But yeah, pretty interesting.
267 00:42:55.970 ⇒ 00:42:59.430 Uttam Kumaran: Yeah, I think you’ll find that it’s always easier to go. More management.
268 00:42:59.810 ⇒ 00:43:06.799 Uttam Kumaran: It’s it’ll be hard, I mean, because I I did this. I went more technical first.st And then I realized that okay.
269 00:43:07.470 ⇒ 00:43:11.529 Uttam Kumaran: in order to really increase your leverage, you have to have a team.
270 00:43:12.860 ⇒ 00:43:13.890 Uttam Kumaran: However.
271 00:43:14.340 ⇒ 00:43:23.680 Uttam Kumaran: you’ll the more you go into business, the less development work you’re gonna be able to do. And so it’s always for me. My advice is to go technical first, st like.
272 00:43:23.860 ⇒ 00:43:29.170 Uttam Kumaran: become as technical as possible, work on the toughest problems, and then
273 00:43:29.610 ⇒ 00:43:41.554 Uttam Kumaran: you can always go into management. People always have a hard time where they go into management first, st and they realize they don’t know anything but lead. They don’t know anything about leading people, and the people that you’re leading will quickly see through that.
274 00:43:42.030 ⇒ 00:43:46.179 Uttam Kumaran: you know. Don’t know. You have no idea what you’re talking about, and so
275 00:43:46.350 ⇒ 00:43:48.980 Uttam Kumaran: I don’t know. That’s my 2 cents on that
276 00:43:52.748 ⇒ 00:44:17.440 Uttam Kumaran: cool. So yeah, I think this is a really great conversation. Let me catch up again with Robert and see sort of what our availability is with clients. Typically, the way we work is we try to bring on people slow, like 10 to 20 h, and sort of give you a chance to take on a couple of tasks for a client that gives you a great opportunity to see what it’s like to work with us. That gives us an opportunity sort of value where you would best fit
277 00:44:18.149 ⇒ 00:44:28.369 Uttam Kumaran: and then we sort of try to move people up from there. So let me catch up again with Robert, and sort of give him a little bit of sense of what we talked about, and and find out if there’s a client that we have right now. That would be a good fit
278 00:44:28.881 ⇒ 00:44:35.920 Uttam Kumaran: but regardless if I can be helpful in any other way, please let me know. And yeah, hope to, you know. Catch up again later this week.
279 00:44:36.980 ⇒ 00:44:39.670 Melina Tsai: Thank you so much for taking the time to talk to me.
280 00:44:39.670 ⇒ 00:44:40.510 Uttam Kumaran: Of course.
281 00:44:40.510 ⇒ 00:44:40.840 Melina Tsai: Really cool.
282 00:44:40.930 ⇒ 00:44:45.190 Uttam Kumaran: Definitely. Okay. Well, have a good week. Yeah. Talk to you soon.