Meeting Title: Brainforge Data Analyst Interview Date: 2026-03-03 Meeting participants: Asad Saleem, Awaish Kumar
WEBVTT
1 00:02:16.460 ⇒ 00:02:16.960 Asad Saleem: Hello?
2 00:02:16.960 ⇒ 00:02:17.560 Awaish Kumar: Hello.
3 00:02:23.120 ⇒ 00:02:24.240 Awaish Kumar: Can you hear me?
4 00:02:24.470 ⇒ 00:02:25.340 Asad Saleem: Yes, yes.
5 00:02:25.930 ⇒ 00:02:27.370 Awaish Kumar: Okay, how are you doing?
6 00:02:27.760 ⇒ 00:02:29.980 Asad Saleem: All good. I hope you’re fine as well.
7 00:02:31.350 ⇒ 00:02:33.480 Awaish Kumar: Okay, yeah,
8 00:02:33.580 ⇒ 00:02:43.610 Awaish Kumar: In this interview, like, we are just going to talk a little bit about Brainforge, and after that, we are going to have a brief introduction.
9 00:02:44.820 ⇒ 00:02:49.900 Awaish Kumar: From you, and then we are going to deep dive into, kind of,
10 00:02:50.290 ⇒ 00:02:56.259 Awaish Kumar: the experience you have, and the kind of projects you have done in the past. So… Sure.
11 00:02:56.380 ⇒ 00:03:01.069 Awaish Kumar: For me, I’m kind of leading data engineering team here at Brainforge.
12 00:03:01.200 ⇒ 00:03:11.310 Awaish Kumar: And, we… Brainforge is a kind of, consulting company. We provide data and AI services to,
13 00:03:11.580 ⇒ 00:03:15.979 Awaish Kumar: mid to large scale enterprises, in the US.
14 00:03:16.240 ⇒ 00:03:25.190 Awaish Kumar: And currently, most of our clients are in the US, but we are working remotely, and our employees are… are, like, all over the…
15 00:03:25.340 ⇒ 00:03:28.390 Awaish Kumar: the world, so a lot from Europe.
16 00:03:28.680 ⇒ 00:03:34.440 Awaish Kumar: And a lot of people from Philippine, India, Pakistan, And,
17 00:03:34.970 ⇒ 00:03:37.950 Awaish Kumar: Yeah, quite a few from US as well.
18 00:03:38.220 ⇒ 00:03:40.600 Awaish Kumar: So, yeah, that’s the…
19 00:03:41.100 ⇒ 00:03:47.680 Awaish Kumar: Yeah, that’s basically it. We can… I can answer if you have any questions at the end of the interview.
20 00:03:48.090 ⇒ 00:03:51.400 Awaish Kumar: Yeah, let’s start with your introduction.
21 00:03:52.190 ⇒ 00:03:58.180 Asad Saleem: Sure. So, I started my career in Affinity as a… as a data engineer, actually.
22 00:03:58.360 ⇒ 00:04:02.759 Asad Saleem: But then I quickly transitioned to an analyst. I figured that’s where I’m more suited.
23 00:04:02.930 ⇒ 00:04:11.759 Asad Saleem: So I have experience on both hands, as well as recently, now I’m leading the data team by interacting with the client team, so it’s like a bridge between
24 00:04:11.880 ⇒ 00:04:31.199 Asad Saleem: AI team and the data team. And just like your company, we also… Affinity also offers AI services, too. We have only large-scale accounts, like, I’m currently leading Virgin Media and O2 in the UK region. Then we also have AT&T. We used to have Verizon, so all these big names in the telco industry.
25 00:04:31.660 ⇒ 00:04:41.510 Asad Saleem: So, yeah, it’s been 5 plus years now, and I saw a brain for which I reached out to Utman. I hope I’m saying the name correctly, and this is Mr. CEO.
26 00:04:41.870 ⇒ 00:04:45.659 Asad Saleem: So, and I told him I’m also looking at a swap, I think…
27 00:04:45.860 ⇒ 00:04:51.739 Asad Saleem: Being in a corporate-type environment slows you down, and I’m looking for a startup to swap as well.
28 00:04:52.090 ⇒ 00:04:53.090 Asad Saleem: So, yeah.
29 00:04:53.350 ⇒ 00:04:58.640 Awaish Kumar: Okay, so what are you looking to… like, what are you looking for in a new role?
30 00:04:59.730 ⇒ 00:05:07.179 Asad Saleem: So, I’m trying to transition to a more client-facing role, but obviously, I prefer being hands-on in terms of analysis and everything.
31 00:05:07.220 ⇒ 00:05:24.450 Asad Saleem: So I think it would be similar, where we have a product, I have the full knowledge of the product, so they rely on me whenever we have to introduce a product to the client, or they want to… so basically, we have to show them value our product is delivering, right? And for that, we need data and through analysis there to convince them, and that’s where I come in.
32 00:05:24.450 ⇒ 00:05:34.209 Asad Saleem: So I’m the gap between stakeholders and our… our own AI team, where our AI team generates the value through models, and I have to convince them that this is the value we have generated.
33 00:05:35.340 ⇒ 00:05:38.890 Awaish Kumar: So, like, yeah, like,
34 00:05:39.130 ⇒ 00:05:42.919 Awaish Kumar: I would like to understand it a bit more in detail on what…
35 00:05:43.120 ⇒ 00:05:47.470 Awaish Kumar: How, like, how hands-on you are with, kind of,
36 00:05:47.990 ⇒ 00:05:56.619 Awaish Kumar: while doing your analysis, so, like… Sure. Do you prefer being client-facing doing analysis work, or do you prefer…
37 00:05:56.730 ⇒ 00:06:02.140 Awaish Kumar: kind of writing DVD models, or doing… writing pipelines, and all of that.
38 00:06:02.830 ⇒ 00:06:13.180 Asad Saleem: Yeah, so as I said, I’m trying to transition out of DE and more of a client-facing role. So initially, the tech stack here was, if you need to, as you said, dbt models, we had talent, yeah.
39 00:06:14.310 ⇒ 00:06:26.140 Awaish Kumar: Yeah, I get it, like, you just transitioned, but that’s, like, you transitioned because the company wanted you to be there, or, like, it’s your personal choice to…
40 00:06:26.480 ⇒ 00:06:28.999 Awaish Kumar: To be able to be as a data analyst.
41 00:06:29.590 ⇒ 00:06:41.610 Asad Saleem: So, as a DE, I was on the back end. So, as I grew more in the more senior role, they put me in the front as well, and now I enjoy being there as well. So, yeah, it’s a personal choice, and it was a career path as well.
42 00:06:41.950 ⇒ 00:06:46.809 Asad Saleem: So initially, when I started off, my tech stack was Talon, SQL, Python.
43 00:06:47.000 ⇒ 00:07:00.279 Asad Saleem: And we… I had to build pipelines, so we fetch clients’ data, and they have massive data sets, so we bring it in our on-prem environment, because that’s where we run the models, not on cloud. Clouds are expensive for us.
44 00:07:00.710 ⇒ 00:07:10.050 Asad Saleem: And so my job was to fetch… build a pipeline that fetches the data daily. Sources could be anything. They had data lakes, some had Oracle, some was manual SFTP files.
45 00:07:10.160 ⇒ 00:07:14.660 Asad Saleem: And then build a pipeline that runs each day from scratch, and then
46 00:07:14.920 ⇒ 00:07:26.080 Asad Saleem: fetch the raw data, convert it in PostgreSQL into features that AI prefers, and then this is… will be the daily pipeline. And the other part was.
47 00:07:26.650 ⇒ 00:07:34.550 Asad Saleem: So, using that data, I have to do analysis that how are we performing? We have to report the invoice and everything. What’s the churn rate?
48 00:07:34.730 ⇒ 00:07:39.970 Asad Saleem: In all the telco KPS, the KPS that we have, yeah.
49 00:07:40.760 ⇒ 00:07:45.660 Awaish Kumar: Okay, I get it. So, like, recently, you have been doing some analysis.
50 00:07:45.790 ⇒ 00:07:51.340 Awaish Kumar: So, like, if you can share a little bit about any of the recent projects.
51 00:07:51.870 ⇒ 00:07:57.659 Awaish Kumar: How it went from, requirements gathering to… Building…
52 00:07:57.920 ⇒ 00:08:08.799 Awaish Kumar: And, like, requirements gathering to finally showing some analysis, but also what delivery will look like, how you shared your…
53 00:08:08.930 ⇒ 00:08:22.669 Awaish Kumar: Findings, and… and then also, if you would like, you can answer if there were any, conflict, like, disagreements between you and the stakeholders, how then you address that.
54 00:08:23.500 ⇒ 00:08:24.140 Asad Saleem: Sure.
55 00:08:24.270 ⇒ 00:08:32.500 Asad Saleem: So, initially, I was on Claro account. I know it’s a big account, but in terms of revenue that we generate from it, it was around $200K for us.
56 00:08:32.720 ⇒ 00:08:36.539 Asad Saleem: So they’ve moved me to version VNO2.
57 00:08:36.740 ⇒ 00:08:53.049 Asad Saleem: And they wanted me to leave that, because there was a lot of changes, planned for that. So, starting with the first change was they were changing their switch, so it’s like the… when you dial in, you can either have Cisco, Genesis, or Raya, these are the major ones, but they were swapping out Amazon Connect.
58 00:08:53.230 ⇒ 00:09:05.859 Asad Saleem: So the entire design based on call data was about to change, and that was the first one. We had to swap our entire pipeline from Cisco to Amazon Connect, for Virgin Media, and then for O2 from Genesis to Amazon Connect.
59 00:09:06.090 ⇒ 00:09:22.849 Asad Saleem: At the same time, I realized that their database is using MySQL, and that’s really slow. It doesn’t have parallel processing, so I told them that we should stop to PostgreSQL, and because we’re not on GCP, so on-prem solution is you use Greenplum, it’s like a distributed clustering.
60 00:09:23.010 ⇒ 00:09:27.319 Asad Saleem: So, the language is PostgreSQL, but then you can apply clustering.
61 00:09:27.660 ⇒ 00:09:35.320 Asad Saleem: And that’s what Greenpoint supports, or CloudBerry, or similar other tools. So, one migration was this switch.
62 00:09:35.520 ⇒ 00:09:52.920 Asad Saleem: The other was database migration, and the third was we’ve had to move our on-prem from the Docklin servers. We had to migrate to the Nosely servers. It’s basically the location where our data centers were hosted. So we had to do 3 parallel migrations at once. They gave me a team of four people.
63 00:09:53.090 ⇒ 00:10:12.090 Asad Saleem: And we had to deploy this hands-on. Obviously, there was a challenge that you… we can’t go offline. We need to remain in production. So we had to maintain the pipelines, the existing ones, which we can call legacy, and at the same time, we have to swap over seamlessly to the new ones. So we had to remain in production on both of them at the same time.
64 00:10:12.200 ⇒ 00:10:16.380 Asad Saleem: So it took us about 3 months, 4 months. The expectation was it would take 6.
65 00:10:16.930 ⇒ 00:10:19.410 Asad Saleem: So this is more on the data engineering front, I would say.
66 00:10:19.560 ⇒ 00:10:22.109 Asad Saleem: Then, at the same time, they were undergoing design.
67 00:10:22.340 ⇒ 00:10:28.409 Asad Saleem: So we were shifting our metric, the optimization metric, so that’s basically to measure the value that we deliver.
68 00:10:28.510 ⇒ 00:10:31.920 Asad Saleem: before it was an on-the-call metric, so if I explain a bit…
69 00:10:32.160 ⇒ 00:10:37.549 Asad Saleem: An agent and customer have an interaction, and let’s say the customer says, I want to have a sale.
70 00:10:37.680 ⇒ 00:10:44.400 Asad Saleem: Whatever the agent marks on the call, that’s the finance sales. But that’s not actually how what happens in reality.
71 00:10:44.640 ⇒ 00:11:04.510 Asad Saleem: So, you could call back afterwards, or you could change your mind, and you could change your outcome. So, rather than mirroring on the call, we now have to look 60 days after the call that what happened, and that’s going to be our new optimization metric. And that’s where my, basically, growth started to happen. The stakeholders, they involved me with the Vision Media clients directly.
72 00:11:04.710 ⇒ 00:11:09.370 Asad Saleem: Their client services team, they want to change the metric, and we need you to build this out.
73 00:11:09.430 ⇒ 00:11:26.979 Asad Saleem: So, in terms of building, I told them this is the data that we need. Currently, you’re only giving us the outcomes that are joined on the call. Now we need the entire database. We need to know what happened 6 years after the call, so we need your CRM as well. We need your transactions as well. So all these outcomes are our data requirements.
74 00:11:27.020 ⇒ 00:11:31.859 Asad Saleem: Then, within that subset, I had calls with the data team, that what exactly do you need?
75 00:11:32.100 ⇒ 00:11:38.059 Asad Saleem: And then I also have to represent our AI team, so they want specific features that helps them in modeling, so this was on the data…
76 00:11:38.180 ⇒ 00:11:41.299 Asad Saleem: Requirements part, then…
77 00:11:41.840 ⇒ 00:11:55.809 Asad Saleem: After I received the data on GCS, so we swap from SFTP to GCS now, so they drop files in GCS, and then we fetch them, and we build our pipeline. So this time I use Airflow. We want to decommission talent, so for orchestration, I use Airflow.
78 00:11:55.820 ⇒ 00:12:06.229 Asad Saleem: And then, the DB was 4Q SQL, Postgres SQL. I prefer doing most of my work on SQL, unless I have to use Python. Otherwise, I prefer SQL.
79 00:12:06.820 ⇒ 00:12:12.839 Asad Saleem: So, searching all the data, building all the optimization metrics, and everything, so…
80 00:12:12.950 ⇒ 00:12:16.160 Asad Saleem: One thing we recently had was we had a conflict.
81 00:12:16.500 ⇒ 00:12:26.710 Asad Saleem: With the O2 team. They came to us and said, you know what, we don’t believe in the value that you’re delivering. We have done some analysis on our end, which is like,
82 00:12:27.050 ⇒ 00:12:30.210 Asad Saleem: We call it LTV, Lifetime Value Analysis of a Customer.
83 00:12:30.480 ⇒ 00:12:34.509 Asad Saleem: And by that, we see that the value you deliver quickly fades away.
84 00:12:34.730 ⇒ 00:12:47.659 Asad Saleem: So unless you show us otherwise, you’re going to shut us down. And that’s where I got my most recognition from. They brought me and said, you know what, you’re going to get shut down unless we convince them otherwise. So I aligned with our director of AI,
85 00:12:47.720 ⇒ 00:12:56.609 Asad Saleem: And we ran our version of the analysis, where we look at the customer value 6 months after the call, 180 days after the call.
86 00:12:56.700 ⇒ 00:13:00.810 Asad Saleem: Just to convince them that, you know what, the value that we deliver does take, and does hold.
87 00:13:01.060 ⇒ 00:13:05.390 Asad Saleem: And we went to the commercial team, that, you know what, this is the value we have generated.
88 00:13:05.560 ⇒ 00:13:07.030 Asad Saleem: And it does hold.
89 00:13:07.180 ⇒ 00:13:15.279 Asad Saleem: 60 days after the call, 180 days after the call, you can see that the values will persist if you look at a subset of the customers as a cohort analysis.
90 00:13:15.450 ⇒ 00:13:18.160 Asad Saleem: And that’s where they gave us a timeline, you know what?
91 00:13:18.740 ⇒ 00:13:25.569 Asad Saleem: Okay, we believe that you are delivering us value, but we want a discount, we want to change the metric.
92 00:13:26.310 ⇒ 00:13:31.410 Asad Saleem: And then… so I would… we want to change this, then we can stay live.
93 00:13:31.540 ⇒ 00:13:37.370 Asad Saleem: So, using the data that we already have, I propose that rather than shutting down and building something new.
94 00:13:37.410 ⇒ 00:13:53.190 Asad Saleem: Let’s build an entire metric that we’re both happy on, while we build the new metric that you prefer. And that’s what we are doing right now. We are live on the entire metric, we are still generating invoice, and at the same time, I’m working with the O2 client team to build the new metric for them.
95 00:13:57.220 ⇒ 00:13:59.490 Asad Saleem: There are more projects, it depends on what you want to know.
96 00:13:59.490 ⇒ 00:14:00.359 Awaish Kumar: Let me know.
97 00:14:00.530 ⇒ 00:14:11.880 Awaish Kumar: Yeah, I understand the project, but I just want to know, like, since you mentioned that, you are working as a data analysts, so, like, how actually you convinced them?
98 00:14:12.360 ⇒ 00:14:15.379 Awaish Kumar: what was your deliverable? What you showed them, like…
99 00:14:15.540 ⇒ 00:14:20.550 Awaish Kumar: Was, like, how you basically convey your thoughts to the client.
100 00:14:21.290 ⇒ 00:14:28.770 Asad Saleem: Sure. So, right now, we were not on lag metric, we were on the call metric. So, here’s what they did. So…
101 00:14:28.930 ⇒ 00:14:33.939 Asad Saleem: We have a call, so basically what we do as well, first, let me explain that, right?
102 00:14:34.350 ⇒ 00:14:47.570 Asad Saleem: So we have calls coming in, and against those calls, you have agents who mark sales or disconnects, right? So what we do is, 50% of those calls, we will decide that what customer should be paired to what agent based on our AI modeling.
103 00:14:47.900 ⇒ 00:14:59.700 Asad Saleem: And then, on those 50% of the calls, let’s say we deliver 50 sales, as compared to the other 50% of the calls, we had 40 sales. Those 10 additional sales are an uplift. And we take a revenue share of that.
104 00:15:00.160 ⇒ 00:15:05.530 Asad Saleem: Now, what they did, analysis was that for those on percentage of calls where we were live.
105 00:15:05.690 ⇒ 00:15:09.660 Asad Saleem: They looked at those customers, Two years afterwards.
106 00:15:09.870 ⇒ 00:15:20.979 Asad Saleem: Does the value still hold? So, obviously, they did this analysis for 2024, they didn’t have enough data for 2 years, so they used, like, whatever data was available at that point. This was, like, 4 months ago.
107 00:15:21.890 ⇒ 00:15:26.669 Asad Saleem: And they realized that the value we delivered in our on-cycle
108 00:15:26.830 ⇒ 00:15:31.439 Asad Saleem: As compared to the 50% of the remaining calls, that really quickly faded away.
109 00:15:32.050 ⇒ 00:15:40.309 Asad Saleem: We were taking a… we told them that the value would stick for 18 months, but by their analysis, it was only sticking on average for 12 months.
110 00:15:40.530 ⇒ 00:15:41.270 Asad Saleem: So…
111 00:15:41.300 ⇒ 00:15:57.619 Awaish Kumar: Okay, I just wanted to know, like, how do you communicate with non-technical stakeholders? For example, while building the presentations, dashboards, were you building any BI dashboards, or your Google Sheets, like…
112 00:15:57.880 ⇒ 00:16:12.019 Awaish Kumar: The number… obviously, you were sharing some numbers, but, like, the… what was the way of, like, just a Slack message, or was it an executive-level, spread DAC, or how it was conveyed?
113 00:16:12.560 ⇒ 00:16:25.980 Asad Saleem: So internally, I do all the analysis, in… and then I… in my SQL, and then I take out the stats in Excel. So, to my own team, I show that. But then to the stakeholders, and then when presenting to the client, I take it out in PPT.
114 00:16:25.980 ⇒ 00:16:36.150 Asad Saleem: We don’t build dashboards because these are time-intensive analysis, and we are at the risk of being shut down, so we take… finalize all the stats and the summaries in a PPT, and that’s where we present with them.
115 00:16:37.080 ⇒ 00:16:44.770 Awaish Kumar: Okay, and then, like, was it, like, the one-time exercise?
116 00:16:45.150 ⇒ 00:16:45.820 Asad Saleem: Nope.
117 00:16:46.160 ⇒ 00:16:48.030 Asad Saleem: Okay. Multiple calls.
118 00:16:48.030 ⇒ 00:16:58.869 Awaish Kumar: Yeah, my question… no, I mean, obviously you are sharing this number after every week or something, like, okay, what does the uplift look like, or how we are doing?
119 00:16:58.980 ⇒ 00:16:59.590 Awaish Kumar: So…
120 00:17:00.320 ⇒ 00:17:12.490 Awaish Kumar: and you are building, as you mentioned, that, because it’s time-consuming, you are not building a dashboard. Like, like, there’s another thought that, okay, we are doing similar exercise.
121 00:17:13.020 ⇒ 00:17:21.569 Awaish Kumar: Every week, repeatedly, like, why not just build some automation and build some dashboard which shows the values as it arrives?
122 00:17:22.079 ⇒ 00:17:41.349 Asad Saleem: Definitely. So… so let’s say monthly invoicing, right? Or some analysis we call MBR, monthly business review. We have a set of, dashboards internally for that. We use Power BI, but that’s a separate team that does it. They use the data that I build them, and then on top of it, they have a dashboard. For those weekly or monthly cadences, we do have dashboards set up.
123 00:17:41.429 ⇒ 00:17:46.609 Asad Saleem: And VM02 plans to build it in a GCP on TabView, and they want my help for that.
124 00:17:46.989 ⇒ 00:18:01.449 Asad Saleem: So, yes, for those stuff, definitely. But then this… something comes up all of a sudden, and the stakeholder says, you want to shut us down? Then we have to do it, it’s time intensive, we have… we get a short time stop, like, one week, come back with your analysis in one week or a few days.
125 00:18:03.079 ⇒ 00:18:04.879 Asad Saleem: So for that, just BPDs.
126 00:18:06.300 ⇒ 00:18:10.179 Awaish Kumar: So… Do you have experience building dashboards?
127 00:18:11.190 ⇒ 00:18:21.509 Asad Saleem: Yeah, I had a project where I was supposed to deploy a generic script that would run on all accounts, and I built that on Superset Apache, because it’s open source.
128 00:18:21.790 ⇒ 00:18:29.489 Asad Saleem: And then I deployed it on AT&T, I deployed on VMO2, Claro, Mech Kable, and multiple other accounts.
129 00:18:29.990 ⇒ 00:18:32.860 Asad Saleem: I don’t prefer… I don’t think I’m an expert.
130 00:18:32.960 ⇒ 00:18:34.390 Asad Saleem: of BI.
131 00:18:34.740 ⇒ 00:18:45.170 Asad Saleem: But if I have to, I can get by. I know it’s weird because I prefer being an analyst, but I’m more hands-on on the raw data itself and interacting with the client for the analysis.
132 00:18:46.060 ⇒ 00:18:51.689 Awaish Kumar: I… yeah, some… it is kind of, something need, like…
133 00:18:51.800 ⇒ 00:19:00.319 Awaish Kumar: There’s a lot of times when the analyst just has to go in and do analysis and share insights, but there are quite a… like, this is also,
134 00:19:00.550 ⇒ 00:19:06.150 Awaish Kumar: Something which data analyst has to do, in their job, like in the dashboards.
135 00:19:07.490 ⇒ 00:19:19.360 Awaish Kumar: So, like, that… like, that… that is something that we… that’s why we are… I asked that question. If you have any experience with any of the tools that, like, previously, apart from Superset.
136 00:19:20.680 ⇒ 00:19:32.179 Asad Saleem: hands-on superset, Power BI, I collaborated, but I wasn’t the owner of the project, and Tableau, I’m going to do it on VMO2 and GCB, but I don’t have hands-on Tableau yet, so Power BI,
137 00:19:32.270 ⇒ 00:19:42.649 Asad Saleem: And Supers said, yes, these are, like, my main dashboarding skills for now. Open to learn, but so far, they don’t require me to build dashboards, they prefer me being more hands-on.
138 00:19:44.210 ⇒ 00:19:51.329 Awaish Kumar: Okay, yeah, that’s okay. So how would you, for example.
139 00:19:51.950 ⇒ 00:19:55.119 Awaish Kumar: If there is a disagreement, between…
140 00:19:55.760 ⇒ 00:19:58.740 Awaish Kumar: Your… between your team, like.
141 00:19:59.450 ⇒ 00:20:06.400 Awaish Kumar: Among your team, so you… you have… For example, delivered some…
142 00:20:07.410 ⇒ 00:20:13.430 Awaish Kumar: Like, you are showcasing some roadmap for… for a… Like, to solve a problem.
143 00:20:13.550 ⇒ 00:20:23.890 Awaish Kumar: And someone disagrees, internally, like, within your team. So how would your, like,
144 00:20:25.350 ⇒ 00:20:28.780 Awaish Kumar: How would you take their feedback and then try to resolve that?
145 00:20:29.110 ⇒ 00:20:31.699 Awaish Kumar: What would be your process? Sure.
146 00:20:32.510 ⇒ 00:20:36.099 Asad Saleem: Okay, so from experience, we have that issue right now.
147 00:20:36.380 ⇒ 00:20:39.559 Asad Saleem: The invoice we are generating for O2, right?
148 00:20:39.990 ⇒ 00:20:48.129 Asad Saleem: So there’s a problem, right? Let’s say someone calls in, and let’s say you want to disconnect your services. You would be sent to the disconnect queue.
149 00:20:48.240 ⇒ 00:21:02.169 Asad Saleem: But on that disconnect queue, if you change your mind, or the agent is so good that he convinces you to re-contract on disconnect rather than disconnecting, so this outcome is not going to be counted currently, because your intent was that you wanted to disconnect.
150 00:21:02.170 ⇒ 00:21:09.000 Asad Saleem: And the same is the opposite. You went to have a sale or a re-contract, but you weren’t happy with the prices, and you ended up disconnected.
151 00:21:09.110 ⇒ 00:21:13.610 Asad Saleem: So if you have an outcome that’s opposite of your intent, we don’t count that outcome right now.
152 00:21:13.830 ⇒ 00:21:16.250 Asad Saleem: And when we generate this invoice.
153 00:21:16.520 ⇒ 00:21:20.080 Asad Saleem: It turned out to be really good. They were giving a 5% uplift.
154 00:21:20.360 ⇒ 00:21:22.310 Asad Saleem: And our AI team was really happy.
155 00:21:22.440 ⇒ 00:21:25.520 Asad Saleem: But then I removed this filter, this, closet.
156 00:21:26.150 ⇒ 00:21:32.879 Asad Saleem: Rather than limiting, I counted all the outcomes in all the queues. And it ended up that we were negative, we would… we would, like.
157 00:21:33.100 ⇒ 00:21:42.450 Asad Saleem: actually costing them 1%, rather than giving a 5% uplift. So I told my AI team that I don’t think this is working out. Right now, you might invoice them.
158 00:21:42.610 ⇒ 00:22:01.840 Asad Saleem: which… and you might get away with it, but as per contract, but then we won’t… I won’t have a good relationship with the client, so because I have to communicate with them, and if I tell them, hey, we have donated you 5% uplift, and later on they check the data by themselves, and they move this filter, they would have a minus 1 person, and we’d be cautioning them, they would definitely shut us down.
159 00:22:01.980 ⇒ 00:22:10.200 Asad Saleem: And this is an ongoing conversation right now. So why your team says, why should we take the burden of that one person? We will get bad reviews that you’re not performing.
160 00:22:10.380 ⇒ 00:22:12.759 Asad Saleem: So, we don’t want to do it.
161 00:22:12.940 ⇒ 00:22:27.030 Asad Saleem: So then I brought in the stakeholders, you know what? There’s a better way to solve this. I’m not telling… I’m… I think, rather than going to the client first and telling them everything, let’s try to solve it by ourselves. So adapt our models, where you take in all the data and train on that.
162 00:22:27.230 ⇒ 00:22:42.059 Asad Saleem: So, you perform well with or without the filters. And eventually, in a few months, your models will be trained to perform with or without the filters anyway, and even if they do check, so it will only be that one month where we perform super bad, but the coming months will be good.
163 00:22:42.230 ⇒ 00:22:44.600 Asad Saleem: So this is an option that I presented them.
164 00:22:44.710 ⇒ 00:22:49.020 Asad Saleem: And I hope the client doesn’t look until, like, a few months from now, but…
165 00:22:49.240 ⇒ 00:22:57.679 Asad Saleem: If AI doesn’t change the models, then I will ask the stakeholder team that we should either penalize our AI team and let the client know that we are not performing right now.
166 00:22:58.900 ⇒ 00:23:00.030 Asad Saleem: Client first.
167 00:23:01.580 ⇒ 00:23:06.270 Awaish Kumar: So, yeah, what was your AI team… did your AI team agree on that?
168 00:23:07.590 ⇒ 00:23:14.159 Asad Saleem: They are happy, and they’ve agreed that, okay, we’ll shift our models, but don’t penalize us for this month.
169 00:23:14.280 ⇒ 00:23:19.789 Asad Saleem: let it be 5%, but we are going to shift from modeling strategy for going forward. That’s the agreement right now.
170 00:23:20.400 ⇒ 00:23:30.450 Asad Saleem: And it’s not my call to make that, whether we take it to the client or not, but it was my call to highlight it, that they might remove these filters and check their invoice without it as well.
171 00:23:32.590 ⇒ 00:23:33.320 Awaish Kumar: Okay.
172 00:23:33.430 ⇒ 00:23:40.949 Awaish Kumar: I think, yeah, I think that’s it from my side,
173 00:23:41.520 ⇒ 00:23:48.589 Awaish Kumar: Yeah, just one more question. Yeah, so is there anything that you had to learn, during the job?
174 00:23:49.290 ⇒ 00:23:53.090 Awaish Kumar: And, basically deliver… Deliver the value.
175 00:23:54.200 ⇒ 00:24:06.400 Asad Saleem: Yeah, everything, bro. So when they threw me in the data engineering, I had to learn Airflow and talent from scratch, and then I ended up on Claro was my fourth account, and I built the pipeline from scratch myself.
176 00:24:06.680 ⇒ 00:24:17.850 Asad Saleem: And then when they threw me in the analysis part, and the building the optimization metrics and everything, I had to learn about client business and more about the switches and how the industry works.
177 00:24:17.960 ⇒ 00:24:22.950 Asad Saleem: And so it’s been a learning experience throughout, and right now I’m learning how to be more client-facing.
178 00:24:23.230 ⇒ 00:24:26.349 Asad Saleem: It’s one thing to have the data on your hand.
179 00:24:26.500 ⇒ 00:24:41.849 Asad Saleem: But you have to… you have… you now have different stakeholders that you have to represent. You want to get the right data for your AI team, so you need to negotiate that. They don’t want to hand over all the data to you. So how do you do it politely, without becoming, like… you can’t really demand from the client?
180 00:24:41.910 ⇒ 00:24:50.269 Asad Saleem: At the same time, you want to make sure that your client is happy, so keep a check on your AI team that they’re not creating some sort of bias as they were right now.
181 00:24:50.660 ⇒ 00:24:57.840 Asad Saleem: And I guess the biggest one that I’m learning right now is being in a call with the stakeholders, not everyone is technical.
182 00:24:58.100 ⇒ 00:25:03.829 Asad Saleem: So I have to stop using technical terms and explain them everything. It might, in my head, it’s pretty clear.
183 00:25:04.070 ⇒ 00:25:08.520 Asad Saleem: But to them, if I look from their perspective, they don’t really understand what I’m saying.
184 00:25:08.640 ⇒ 00:25:18.500 Asad Saleem: GCP, what is GCS, what are the transformations that I’m making on top of what are these filters? Not everyone understands those parts, so I have to explain them.
185 00:25:18.810 ⇒ 00:25:20.150 Asad Saleem: In a genetic form.
186 00:25:21.300 ⇒ 00:25:24.819 Asad Saleem: So it’s been a learning experience throughout, and as you mentioned, like.
187 00:25:24.930 ⇒ 00:25:37.500 Asad Saleem: Dashboarding, I haven’t been required since that one project that I did, but if I have to, I would love to learn it. And if I’m needed to understand the product more and be more hands-on analytical front, that’s fine as well.
188 00:25:39.110 ⇒ 00:25:40.460 Awaish Kumar: Okay.
189 00:25:40.740 ⇒ 00:25:42.660 Awaish Kumar: And like…
190 00:25:42.790 ⇒ 00:25:49.180 Awaish Kumar: My question would be, then, when you were… when you tried to learn Airflow, so, like.
191 00:25:49.340 ⇒ 00:25:57.849 Awaish Kumar: What… what that looked like? How… what was the project? What… how the deadline looked like? Were you the only…
192 00:25:58.130 ⇒ 00:26:02.609 Awaish Kumar: Honor of the project, or is what it… like, you’re working in a team?
193 00:26:03.770 ⇒ 00:26:16.250 Asad Saleem: So when I learned talent, I learned it from scratch, everything hands-on. I was doing it independently. Now, because my expectations are a lot different, I was a lead, I had a team of four, so I delegated the airflow to one of my teammates.
194 00:26:16.280 ⇒ 00:26:31.999 Asad Saleem: And then I delegated the Green Plum migration from MySQL to the other teammate. So I tried to hand over the other projects, because one of my OKRs, or you can say the evaluation for the next term was that you have to delegate more. You can’t try to do everything by yourself.
195 00:26:32.080 ⇒ 00:26:41.919 Asad Saleem: And that was, like, the biggest concern for me. I was like, if I don’t do it hands-on, how would I learn? And he said, you know what, you’ll get time to learn, but you have to look at the bigger picture, what the client wants.
196 00:26:42.020 ⇒ 00:26:51.950 Asad Saleem: So you’re responsible for what the client wants. In terms of hands-on, you can always go back to your team and ask them, okay, how did you do it? And you can keep a check on then an oversight. So that’s what I did for Airflow.
197 00:26:52.290 ⇒ 00:27:00.569 Asad Saleem: So if someone asks me to build a pipeline from scratch in Telen, I can do it. But for Airflow, I’ll definitely need some help. But now you have ChatGPT, so it’s a lot easier.
198 00:27:02.690 ⇒ 00:27:07.250 Awaish Kumar: Okay, I think that’s it, from my side, like, if you have any other questions…
199 00:27:07.710 ⇒ 00:27:09.070 Awaish Kumar: We’d be happy to answer.
200 00:27:09.790 ⇒ 00:27:17.409 Asad Saleem: Sure. So, it’s a bit different interview process. You weren’t asking me technical questions, which I’m fine with that as well.
201 00:27:17.520 ⇒ 00:27:24.930 Asad Saleem: Like, for example, you didn’t… people usually ask, like, can you solve this problem, or XYZ query, or XYZ code?
202 00:27:25.180 ⇒ 00:27:30.889 Asad Saleem: I’m happy with that, but what are the… what are you looking for in person when you’re hiring?
203 00:27:31.400 ⇒ 00:27:36.670 Asad Saleem: Are you looking more someone who builds dashboard? That’s what I got the impression from. Are you looking for more client-facing?
204 00:27:37.370 ⇒ 00:27:51.530 Awaish Kumar: So, as I mentioned, we are a data consultancy company, and we are kindly open to hire for any of the roles. We are hire for hiring for all data streams, data analyst, data analytics engineering, and also in the data engineering.
205 00:27:52.030 ⇒ 00:27:53.310 Awaish Kumar: departments.
206 00:27:53.420 ⇒ 00:28:00.300 Awaish Kumar: So… Yeah, that’s why I was trying to understand, instead of going deeper into…
207 00:28:00.680 ⇒ 00:28:06.779 Awaish Kumar: asking about caries or anything, I’m just trying to understand how you work on the project.
208 00:28:07.420 ⇒ 00:28:10.060 Awaish Kumar: What are your, basically, the…
209 00:28:10.660 ⇒ 00:28:15.170 Awaish Kumar: There’s something that you want to do, instead of just, like, obviously,
210 00:28:15.540 ⇒ 00:28:23.350 Awaish Kumar: In a com… in a company, you can be placed anywhere, if needed, but obviously, it’s also…
211 00:28:24.060 ⇒ 00:28:41.900 Awaish Kumar: Like, we also want to make the placement, like, kind of a win-win for both the parties, so that, like, if you are… you want to do data analyst, then you should be there, instead of being, any other… in any other work streams.
212 00:28:42.130 ⇒ 00:28:45.310 Awaish Kumar: Sure. Yeah, that’s one of the things that I wanted…
213 00:28:45.530 ⇒ 00:28:50.860 Awaish Kumar: understand from this interview. So, that’s basically… yeah, that’s basically it.
214 00:28:51.480 ⇒ 00:28:58.549 Asad Saleem: Sure. I guess lastly, if you think there’s some confusion… confusion point that you would like me to clarify.
215 00:29:00.340 ⇒ 00:29:10.959 Awaish Kumar: No, I think I’m good. I already asked my question, so I would have any confusion, I could have asked. But I’m just here, you know, to answer if you have any questions.
216 00:29:11.080 ⇒ 00:29:16.249 Awaish Kumar: About brain folds, how we work, what time zone we go, things like that.
217 00:29:16.850 ⇒ 00:29:29.280 Asad Saleem: Yeah, time zone is fine. I’m already remote work, so I’m flexible. And BrainProach, I did take a look extensively. I’ve been following, Utman, the CEO, I hope I’m saying his name correctly.
218 00:29:29.380 ⇒ 00:29:35.709 Asad Saleem: And I looked at a website. I understand what you guys are doing. It’s a bit similar to what we do.
219 00:29:35.980 ⇒ 00:29:39.559 Asad Saleem: I’m guessing you have more clients on a smaller scale?
220 00:29:40.510 ⇒ 00:29:42.290 Asad Saleem: And you’re trying to expand the base.
221 00:29:44.210 ⇒ 00:29:51.609 Awaish Kumar: Yes, so it’s, like, it’s a startup, so obviously we have clients, but we have, like,
222 00:29:51.960 ⇒ 00:29:59.819 Awaish Kumar: Large enterprises as a client, but we are also continuing to support our smaller clients, which we had earlier.
223 00:29:59.940 ⇒ 00:30:03.220 Awaish Kumar: And we are developing our, yeah, client base.
224 00:30:03.710 ⇒ 00:30:08.180 Asad Saleem: Sure. That’s what I’m looking for. I feel like I’m becoming stagnant. I’m just stuck at…
225 00:30:08.450 ⇒ 00:30:14.629 Asad Saleem: reporting for one company, VM02. And if I look at the next two years, I’m still going to be working for them only.
226 00:30:14.800 ⇒ 00:30:19.259 Asad Saleem: I would rather work in a fast-paced environment where I get more opportunity to learn and grow.
227 00:30:20.280 ⇒ 00:30:24.020 Awaish Kumar: Yeah, since, yeah, that’s one thing that you brought up, so…
228 00:30:24.240 ⇒ 00:30:34.820 Awaish Kumar: Since you are working on a single client, yeah, we have a different, work, strategy, like, each person might be… have… might be working in a…
229 00:30:35.020 ⇒ 00:30:37.400 Awaish Kumar: 2 to 3 clients at the same time.
230 00:30:37.400 ⇒ 00:30:38.050 Asad Saleem: Perfect.
231 00:30:38.060 ⇒ 00:30:42.509 Awaish Kumar: So if you’re, like, I think you would have to do context switching,
232 00:30:42.920 ⇒ 00:30:51.829 Awaish Kumar: quite often, and, also, like, although I mentioned that we have big clients, but we,
233 00:30:52.230 ⇒ 00:31:07.489 Awaish Kumar: like, optimize our resourcing, so we don’t, like, assign one person just on one client. It will be divided based on how much DEA work required for that client, or a data analyst work required for that client.
234 00:31:07.490 ⇒ 00:31:13.419 Awaish Kumar: So that’s why how… that’s how we optimize our resourcing for the client as well, so we can, do an…
235 00:31:13.430 ⇒ 00:31:14.710 Awaish Kumar: the optimal billing.
236 00:31:15.760 ⇒ 00:31:23.059 Asad Saleem: definitely makes a lot of sense to be split. I have been split in multiple accounts before, but yeah, I totally understand.
237 00:31:23.630 ⇒ 00:31:24.600 Asad Saleem: Okay.
238 00:31:24.920 ⇒ 00:31:29.390 Awaish Kumar: Okay, I think then… That’s it, so…
239 00:31:29.550 ⇒ 00:31:36.510 Awaish Kumar: I think, from our, like, the recruiters from our team will be in touch with you for the next steps.
240 00:31:36.770 ⇒ 00:31:39.969 Awaish Kumar: Thank you for the… Thank you for your time, yeah.
241 00:31:40.930 ⇒ 00:31:41.640 Asad Saleem: Thank you, Raish.
242 00:31:41.940 ⇒ 00:31:42.530 Awaish Kumar: Right.