Meeting Title: Brainforge Interview w- Awaish Date: 2026-02-11 Meeting participants: Deepika Sethi, Awaish Kumar
WEBVTT
1 00:09:31.620 ⇒ 00:09:33.249 Awaish Kumar: Hi, the people.
2 00:09:33.250 ⇒ 00:09:35.999 Deepika Sethi: Hi, Avish. It’s Avish, right?
3 00:09:36.250 ⇒ 00:09:36.870 Awaish Kumar: Yep.
4 00:09:37.070 ⇒ 00:09:38.070 Awaish Kumar: How you doing?
5 00:09:38.380 ⇒ 00:09:39.680 Deepika Sethi: I’m good, how are you?
6 00:09:40.990 ⇒ 00:09:42.169 Awaish Kumar: I’m good as well.
7 00:09:42.620 ⇒ 00:09:43.270 Deepika Sethi: Great.
8 00:09:43.590 ⇒ 00:09:48.380 Awaish Kumar: Okay, in this, end call, like, we are just going to…
9 00:09:48.700 ⇒ 00:09:54.300 Awaish Kumar: get to know each other. I’m going to introduce myself and the company and what we do here.
10 00:09:54.650 ⇒ 00:10:03.269 Awaish Kumar: The working style and the culture, and then, like, we are going to get to know a little bit about yourself and your experiences.
11 00:10:03.520 ⇒ 00:10:04.340 Deepika Sethi: Okay, yeah.
12 00:10:04.710 ⇒ 00:10:13.240 Awaish Kumar: Okay. So, yeah, as… my name is Abish Kumar, and I have… I’m the data engineering lead here at BrainFroge.
13 00:10:14.700 ⇒ 00:10:22.410 Awaish Kumar: I, like, I have, like, around 10 years of experience working as a data engineer at multiple startups, growth stage companies, and…
14 00:10:23.350 ⇒ 00:10:26.319 Awaish Kumar: Help them build their data foundations.
15 00:10:26.990 ⇒ 00:10:30.840 Awaish Kumar: Yeah, so that’s basically about me, and then…
16 00:10:30.960 ⇒ 00:10:36.050 Awaish Kumar: For the Brain Forge, we are… our data and AI consultancy
17 00:10:36.220 ⇒ 00:10:46.800 Awaish Kumar: services company operating, like, remotely. Everybody works here, remotely, 100%, and
18 00:10:47.370 ⇒ 00:10:55.290 Awaish Kumar: We have employees from across the world, from… Europe, North America.
19 00:10:55.670 ⇒ 00:11:03.320 Awaish Kumar: and, UAE, India, Pakistan. So, we have… Apart from that,
20 00:11:03.930 ⇒ 00:11:12.180 Awaish Kumar: Yeah, we are basically work with, like, medium to large-scale companies to provide them with,
21 00:11:12.590 ⇒ 00:11:17.889 Awaish Kumar: AI consultancy services, or help them build the data foundations.
22 00:11:18.230 ⇒ 00:11:26.129 Awaish Kumar: That’s mainly about Brainforge. As you must have also seen in the job description, we… the Brainforge can
23 00:11:26.920 ⇒ 00:11:36.520 Awaish Kumar: hire you as a full-time employer, can work in a contracting setting. As everybody is remote, we prefer
24 00:11:37.690 ⇒ 00:11:45.610 Awaish Kumar: A lot of async communication, We focus on documentation. That’s, like, the culture here.
25 00:11:45.830 ⇒ 00:11:50.480 Awaish Kumar: That’s how people normally work. They work in their own style.
26 00:11:50.870 ⇒ 00:11:55.609 Awaish Kumar: And then communicate via, writing documentation.
27 00:11:56.910 ⇒ 00:12:00.750 Awaish Kumar: Okay, now, can you, yeah, can you please introduce yourself?
28 00:12:01.140 ⇒ 00:12:08.420 Deepika Sethi: Sure, so you know my name already. I’m currently pursuing an MS in Information Technology from Fordham University.
29 00:12:08.430 ⇒ 00:12:22.899 Deepika Sethi: And prior to this, I have got around 10 years of experience in varied domains. Last, I was working with London Stock Exchange Group as a product owner in operational resilience, where I was working around two things. One is the theme on data…
30 00:12:22.900 ⇒ 00:12:35.239 Deepika Sethi: management and data governance, where my work was, since we were reporting the operational resilience reports, which are more like, you know, ensuring that your company keeps working, even when there’s, any,
31 00:12:35.350 ⇒ 00:12:38.040 Deepika Sethi: Disruption or anything else in the market.
32 00:12:38.080 ⇒ 00:12:58.719 Deepika Sethi: So my work was more around working on the metadata of the applications and services which LSEC, London Stock Exchange Group provides, so as to make sure that the data is accurate, coming from the authorized data source. And since it was a new team, we had just started building the data management thing. So there were two parts of the team. One was the data governance policy team.
33 00:12:58.720 ⇒ 00:13:00.019 Deepika Sethi: Which would really…
34 00:13:00.020 ⇒ 00:13:11.210 Deepika Sethi: create policies, and our team was… my name was basically involved in enforcing those policies, right? So if the policy says the data has to come from authorized data source.
35 00:13:11.250 ⇒ 00:13:22.650 Deepika Sethi: how will you define the authorized data source, right? So, my work was to work with the users, understand the requirement, and help them really identify the authorized data source, rather than doing
36 00:13:22.650 ⇒ 00:13:45.820 Deepika Sethi: work using manual Excels or something. And then, additionally, once we have defined the authorized data source, we would go into the process workflows and identify the critical data elements. So, as to make sure that our data management is not just derived from the technical side of it, but the business is equally involved in the process, and they have their say, and ultimately, the data delivers the value.
37 00:13:46.090 ⇒ 00:14:02.139 Deepika Sethi: So, it would comprise, we were still in the stage where we were identifying the authorized data sources, and we were… we already had built some pipelines using Snowflake and Apache Kafka, and then this was going back to reporting, reporting tools, but,
38 00:14:02.490 ⇒ 00:14:09.830 Deepika Sethi: We were still looking at the data quality sides of it, ensuring that we understand the data maturity of the organization.
39 00:14:09.850 ⇒ 00:14:22.459 Deepika Sethi: The other part was more towards regulatory traceability. Along with doing the data thing, we were a technical team, so we would keep looking at the opportunities which we could use. We were exploring an opportunity where
40 00:14:22.460 ⇒ 00:14:34.639 Deepika Sethi: instead of manually reading the regulations, you know, from a website, or any single person doing so and missing things, we could really automate this part using AI, where AI could do the web scraping.
41 00:14:34.640 ⇒ 00:14:43.599 Deepika Sethi: identify either the new regulations pertinent to LSEG, or any change in the current regulations, and then
42 00:14:43.600 ⇒ 00:15:01.189 Deepika Sethi: if we could really map it to internal policies and controls using AI. So this was still in, what I will say, in a very initial stage where we were defining the problem statement and the product vision, and doing some prototyping around it. So that was with the London Stock Exchange Group.
43 00:15:01.190 ⇒ 00:15:13.010 Deepika Sethi: Prior to that, I was working mostly in banking domain and regulatory technology. I was with Dodge Bank working on the regulatory report. So how it works is, for every bank, they generally have to report
44 00:15:13.160 ⇒ 00:15:35.230 Deepika Sethi: some numbers in terms of report. It may be your accounting numbers, it may be your transactions to the central bank, or a central authority within those jurisdictions. Plus, there could be additional reports, which are furnished to international organizations. So, I was working on two sides of it. I was working on OTC reports, where my work involved understanding those regulations.
45 00:15:35.230 ⇒ 00:15:39.769 Deepika Sethi: Ensuring that the requirements are clear, the data pipeline is clear.
46 00:15:39.770 ⇒ 00:15:41.619 Deepika Sethi: We started from,
47 00:15:41.650 ⇒ 00:15:48.890 Deepika Sethi: The system was already in place for my work, started from understanding the data required for those reports.
48 00:15:48.920 ⇒ 00:16:07.180 Deepika Sethi: source… ensuring that sourcing is complete. If not, reach back to data sourcing teams and receive the data, then understanding the technical and business validations around it, the entire mapping from staging to centralized database to ultimately where the report is being used.
49 00:16:07.180 ⇒ 00:16:16.560 Deepika Sethi: And work with my data team, product team, and technical team to ensure that reports are very well automated, and ultimately, the generated report is correct.
50 00:16:16.560 ⇒ 00:16:25.740 Deepika Sethi: And, supporting, testing, and UAT, yeah. Prior experience is pretty on same… much on same lines, I think that will be pretty all. If you want, I can go…
51 00:16:26.320 ⇒ 00:16:44.599 Awaish Kumar: we can just go through it, like, step by step. We will get to know all over, like, you’re all over, like, what you have been doing so far. So that’s okay. What I wanted to understand more is, like, when you are working, when you mentioned you were working with governance.
52 00:16:44.710 ⇒ 00:16:51.219 Awaish Kumar: and reporting, and all of that. I want to understand how you were… like, what exactly you were doing, like…
53 00:16:51.330 ⇒ 00:16:54.509 Awaish Kumar: I… what I get from you is, like, you are…
54 00:16:54.870 ⇒ 00:17:01.959 Awaish Kumar: Basically, facilitating… facilitating teams to understand, like, where the authorized data source is.
55 00:17:02.190 ⇒ 00:17:09.400 Awaish Kumar: If the policy has been changed, then, like, what should we do? Like, they’re defining maybe the next steps.
56 00:17:09.460 ⇒ 00:17:22.290 Awaish Kumar: But, in terms of data, were you involved in any kind of, like, data analysis, or reporting, or investigations, and things like that?
57 00:17:22.780 ⇒ 00:17:43.369 Deepika Sethi: Yeah, so basically, it was two parts. One is obviously facilitating, because I was working as a product owner. Second was when I mentioned we were working with the users to understand the process flow. So my work would… we were taking a top-down approach, right? We would first identify the process. Let’s say, if you say, I want to identify my important application.
58 00:17:44.050 ⇒ 00:17:54.649 Deepika Sethi: now we have to dig down, understand the data and the elements which made this application, very important, right? So we will go with the users, try to understand
59 00:17:55.010 ⇒ 00:18:07.460 Deepika Sethi: amongst the entire application, right? What is the data that they are looking at which makes it, an important pet, and an important application? Like, it may be, like, number of customers using those applications, okay?
60 00:18:07.710 ⇒ 00:18:11.070 Deepika Sethi: The monetary impact that a customer will have
61 00:18:11.320 ⇒ 00:18:24.800 Deepika Sethi: if this application goes down. Now, on a high level, it looks like we are looking at the application, but ultimately, everything goes into the data, because it’s the metadata of the app where you define what are the critical elements. So we would sit down with the users.
62 00:18:25.620 ⇒ 00:18:27.420 Deepika Sethi: Take each element.
63 00:18:27.420 ⇒ 00:18:28.489 Awaish Kumar: they use this.
64 00:18:28.920 ⇒ 00:18:31.770 Deepika Sethi: When you say users, does that mean…
65 00:18:32.430 ⇒ 00:18:36.090 Awaish Kumar: Customer, internal stakeholder? Who are you referring to?
66 00:18:36.090 ⇒ 00:18:47.550 Deepika Sethi: It will be majorly internal stakeholders, because ultimately, they were responsible for furnishing the reports to the outside community, and to work with
67 00:18:47.710 ⇒ 00:19:00.099 Deepika Sethi: consumers to ensure that applications are up, so we would work with the internal stakeholders, the operation resilience stakeholders, where we would really spend time with them, understanding the metadata of the app.
68 00:19:00.540 ⇒ 00:19:03.380 Deepika Sethi: Identifying each element which impacts, like.
69 00:19:03.380 ⇒ 00:19:06.349 Awaish Kumar: With the metadata, what exactly we use.
70 00:19:06.350 ⇒ 00:19:06.730 Deepika Sethi: Yeah, I was.
71 00:19:06.730 ⇒ 00:19:07.290 Awaish Kumar: Great hub.
72 00:19:07.530 ⇒ 00:19:22.720 Deepika Sethi: Coming to that. So, obviously, when it comes to data, there has to be a primary key. When I talk about an application, I need to understand whether it’s application ID. So, technically, for me, application ID makes more sense, but I have to understand from them
73 00:19:22.750 ⇒ 00:19:38.159 Deepika Sethi: Is the prime DK should be the application ID or application name. So we’ll start with those very basic elements. Then we go into further details, like, what are the dependency of these applications? It may happen, right? You’re using an application, but it is taking data from multiple sources.
74 00:19:38.160 ⇒ 00:19:54.090 Deepika Sethi: or it is further, providing downstream data to multiple applications, which make it more important, right? So understanding those dependencies and those data elements which are actually linked to, basically, the foreign keys which ultimately, lead to the,
75 00:19:54.090 ⇒ 00:20:02.619 Deepika Sethi: data from other tables, or other applications, or maybe other systems. Understanding those elements. Then there’s something called RTO and RPO.
76 00:20:02.620 ⇒ 00:20:20.949 Deepika Sethi: Which is more about… now, this becomes a very important part of data limits. You can’t really leave it when it… we save those timings in the table, right? RTO means Recovery Time Objective, which is, like, if you are providing service to any outside company, it may be service or an application, you can’t really afford that your system goes down, right?
77 00:20:21.130 ⇒ 00:20:28.130 Deepika Sethi: Ultimate, because that’s where the reliability comes from. So we would work with user to understand those
78 00:20:28.380 ⇒ 00:20:41.870 Deepika Sethi: RTOs, how those should be reported, what should be the, you know, what kind of rules should be there, because these are really important part of it. Like, how to save it, how… what will be the format, yeah.
79 00:20:43.160 ⇒ 00:20:46.690 Awaish Kumar: Just so I understand, so you have an op…
80 00:20:46.880 ⇒ 00:20:56.630 Awaish Kumar: Operational Resiliency team with which you sit and understand the app’s metadata and how it is being collected.
81 00:20:56.880 ⇒ 00:20:57.970 Awaish Kumar: Right? Yeah.
82 00:20:59.750 ⇒ 00:21:05.940 Awaish Kumar: Okay, now that we… you… like, and you… are you part of… That data collection, or…
83 00:21:07.220 ⇒ 00:21:24.479 Deepika Sethi: I’m part of ensuring the data quality on that data, and not really the collection, because collection comes from the sources, front-end teams which, from the applications and sources which front-end team is using. I’m… I come more from the back… back-end team of it, where once we have received it.
84 00:21:25.180 ⇒ 00:21:39.870 Awaish Kumar: Let’s frame it like… so, once, our product team, for example, they are capturing those events, and they are sending data to the backend, you sit with the operational resilience team and figure out
85 00:21:40.210 ⇒ 00:21:45.310 Awaish Kumar: If that data That is captured, makes sense. Is that… Yeah.
86 00:21:45.940 ⇒ 00:21:58.210 Deepika Sethi: Yeah, so we would identify the… Now, obviously, to understand the accuracy, timeliness, and data quality dimensions of this data, we would try to understand the rules. We will do the
87 00:21:58.620 ⇒ 00:22:15.380 Deepika Sethi: checking and the integration, reconciliation of the data to make sure that this is what is expected, and we also try to understand the lineage of the data. Where is it coming from? Where it is supposed to flow? What are the transformations that are supposed to happen on that? So, basically.
88 00:22:15.380 ⇒ 00:22:18.199 Awaish Kumar: How are you, measuring data quality?
89 00:22:18.550 ⇒ 00:22:33.520 Deepika Sethi: So, there were three, multiple parameters, starting with the accuracy. Accuracy means the data is accurate, like, whatever has been sent by the source system is received at, at the backend. So, we would do two kinds, three kinds of stuff, one being
90 00:22:33.670 ⇒ 00:22:44.790 Deepika Sethi: That you count the rows, right? Whatever number of rows were sent by the system, those are saved. Then you do the checksum. This is, like, the technical validation. Then comes the values around it.
91 00:22:44.900 ⇒ 00:22:47.139 Deepika Sethi: So, we would have some rules around it.
92 00:22:47.900 ⇒ 00:22:56.440 Deepika Sethi: Like, ensuring that some value really… percentage can’t be more than 100, right? So we would have those kinds of rules around it, that it doesn’t go beyond.
93 00:22:56.440 ⇒ 00:22:56.980 Awaish Kumar: Yup.
94 00:22:57.260 ⇒ 00:23:09.270 Awaish Kumar: I just… I just want to know, like, I understand that you, based on your domain knowledge, or business domain knowledge, you figured out the threshold that you have set up for
95 00:23:09.320 ⇒ 00:23:19.390 Awaish Kumar: for checking, but I want to more understand more, like, what was the process? How were you checking it? Like, manually? Was it automated? How it was automated?
96 00:23:20.150 ⇒ 00:23:36.239 Deepika Sethi: So, yeah, it was automated. In London Stock Exchange Group, we were using multiple tools around, like, obviously, within Snowflakes, we had some rules defined, which will run automatically every day, and then we would have the dashboards, which would show us the status of those rules.
97 00:23:36.330 ⇒ 00:23:55.499 Deepika Sethi: Additionally, we were kind of using those Kafka pipelines for the streaming of real-time data, but when it comes to Dodge Bank, it was more… since banking is more of a, you know, conservative industry, it was all SQL, so we used to have the SQL tools, and
98 00:23:55.550 ⇒ 00:24:09.150 Deepika Sethi: on… based on that, we used to have those control framework defined. Control framework would have rules for… automated rules for, accuracy, completeness, timeliness, and any error that occur.
99 00:24:09.150 ⇒ 00:24:12.880 Awaish Kumar: for example, you mentioned Snowflake. How…
100 00:24:13.100 ⇒ 00:24:17.639 Awaish Kumar: Were your… how did you define those rules in Snowflake?
101 00:24:18.850 ⇒ 00:24:24.469 Deepika Sethi: I’m sorry if you could, like, explain a bit more how… in terms of how would…
102 00:24:24.470 ⇒ 00:24:30.799 Awaish Kumar: that to, define rules in Snowflake for data quality checks.
103 00:24:32.040 ⇒ 00:24:35.830 Awaish Kumar: That, like, how did you do that? Like, I just want to understand.
104 00:24:36.060 ⇒ 00:24:50.820 Deepika Sethi: So, if I could give you an example from SQL, if that… that’s fine, because I’m more comfortable with that. So, it will be some sort of a batch or SQL script that will run on top of the data. So, I’ll schedule it for every day. So, let’s say my feed is scheduled to come at,
105 00:24:50.820 ⇒ 00:24:57.330 Deepika Sethi: 10 o’clock in the morning. So there will be another batch script that will run on top of it once the data is there, which will go inside the data.
106 00:24:57.330 ⇒ 00:25:00.030 Deepika Sethi: And then run those,
107 00:25:00.190 ⇒ 00:25:06.670 Deepika Sethi: rules around, like, greater than or something, which we will have used. Sorry, Aish, can’t hear you.
108 00:25:07.840 ⇒ 00:25:08.959 Awaish Kumar: Can you hear me now?
109 00:25:08.960 ⇒ 00:25:10.209 Deepika Sethi: Yeah, no, I can hear you.
110 00:25:10.210 ⇒ 00:25:20.470 Awaish Kumar: Yeah, I said… I get it now, I just want to understand that, okay, you have some SQL running for data quality checks, that’s okay, but
111 00:25:20.660 ⇒ 00:25:26.620 Awaish Kumar: Did you… what problems or challenges did you face while doing data quality checks?
112 00:25:27.840 ⇒ 00:25:36.829 Deepika Sethi: When it comes to data quality, I think one of the major challenges which I had seen was manual entry, because on the front end, it’s always a manual entry.
113 00:25:36.940 ⇒ 00:25:40.539 Deepika Sethi: These guys really, obviously, everyone tries their best, but…
114 00:25:40.950 ⇒ 00:25:46.600 Deepika Sethi: That manual error really creates a challenge when things are not really matching.
115 00:25:46.940 ⇒ 00:25:48.449 Deepika Sethi: That’s one part of it.
116 00:25:48.840 ⇒ 00:25:53.450 Deepika Sethi: Other, when it comes to data quality, I would say…
117 00:25:54.210 ⇒ 00:25:57.029 Deepika Sethi: I’m just thinking a bit loud out here.
118 00:25:57.160 ⇒ 00:26:01.689 Deepika Sethi: Based on my experiences. You know, sometimes the data might…
119 00:26:02.160 ⇒ 00:26:11.059 Deepika Sethi: Not really be useful, because it wasn’t given a forethought on how will it be used at the downstream, right?
120 00:26:11.130 ⇒ 00:26:29.429 Deepika Sethi: Well, a very simple example would be a European bank generally either uses GBP or Euro as their base currency, right? But if I have been really getting all the data in USD, and nobody really realized it, the problem lies when nobody really realized that part of it.
121 00:26:29.430 ⇒ 00:26:34.319 Deepika Sethi: And you are doing it further. So, understanding the domain and the,
122 00:26:34.320 ⇒ 00:26:37.980 Deepika Sethi: The nuances of how the data is being used.
123 00:26:38.450 ⇒ 00:26:43.760 Deepika Sethi: that understanding not being there and captured in the system is another challenge, which I… so…
124 00:26:43.860 ⇒ 00:26:46.029 Awaish Kumar: Which I kind of faced.
125 00:26:47.810 ⇒ 00:27:03.220 Awaish Kumar: Yeah, like, that is a data quality issue, like, I get it. So, like, you have data in different currency, and you use it in a, like, in a different, like, with a different definition, obviously, that’s going to be wrong, but…
126 00:27:03.300 ⇒ 00:27:11.410 Awaish Kumar: then I assume that you have already defined some checks for that, so I’m just trying to understand when you have the checks.
127 00:27:11.820 ⇒ 00:27:14.039 Awaish Kumar: And what, like, doesn’t…
128 00:27:14.650 ⇒ 00:27:21.019 Awaish Kumar: Like, you have the checks in place, assuming that. Did you… do you think there is anything else?
129 00:27:21.210 ⇒ 00:27:24.920 Awaish Kumar: Which can be a problematic or a challenge to solve.
130 00:27:26.410 ⇒ 00:27:29.399 Deepika Sethi: Okay, once I have the checks in place.
131 00:27:30.300 ⇒ 00:27:35.490 Awaish Kumar: Yeah, like, obviously, when checks are not in place, that means, like,
132 00:27:35.940 ⇒ 00:27:52.710 Awaish Kumar: Yeah, that’s a different story, right? We don’t have the quadrical quality checks, but when you… for example, I have defined… I know there are 30 core fields for me, which are really critical, and I want to define some checks for that. I have defined those, but now…
133 00:27:53.910 ⇒ 00:27:56.379 Awaish Kumar: Do you think there could be any challenge?
134 00:27:57.320 ⇒ 00:27:58.860 Awaish Kumar: You mean after defining…
135 00:28:00.250 ⇒ 00:28:10.659 Deepika Sethi: Yeah, I really couldn’t recall anything on top of my head, but obviously there could be, challenges around performance of the data.
136 00:28:11.260 ⇒ 00:28:11.940 Deepika Sethi: Y’all.
137 00:28:11.940 ⇒ 00:28:15.940 Awaish Kumar: Okay, have you used dbt?
138 00:28:16.540 ⇒ 00:28:17.230 Deepika Sethi: Sorry?
139 00:28:17.710 ⇒ 00:28:19.260 Awaish Kumar: Have you used dbt?
140 00:28:19.520 ⇒ 00:28:21.610 Deepika Sethi: No, I haven’t worked on DVD yet.
141 00:28:22.290 ⇒ 00:28:25.079 Awaish Kumar: Have you… how would you rate yourself with SQL?
142 00:28:25.590 ⇒ 00:28:31.359 Deepika Sethi: I think SQL, I’m decent since, I’m pretty… I had to extract the data for the…
143 00:28:31.520 ⇒ 00:28:44.160 Deepika Sethi: reconciliation and have a quick understanding of what things look like, so I think… would be around 6 to 7, considering I can do any things around joins, group buy, and CTs.
144 00:28:44.580 ⇒ 00:28:57.279 Awaish Kumar: Okay, that’s good to know. And, like, normally, we have multiple, data streams here at BrainForge. I know that you have applied for a data associate role, that…
145 00:28:57.690 ⇒ 00:29:01.360 Awaish Kumar: It’s more like, Could be…
146 00:29:02.060 ⇒ 00:29:10.439 Awaish Kumar: Like, data… like, you might have… you obviously have to run queries to do the data investigation, but then you might also have to…
147 00:29:11.150 ⇒ 00:29:14.459 Awaish Kumar: To reporting work.
148 00:29:14.780 ⇒ 00:29:18.179 Awaish Kumar: So, how familiar are you with the reporting tools?
149 00:29:18.510 ⇒ 00:29:21.250 Awaish Kumar: BI tools, or Google Sheets.
150 00:29:21.640 ⇒ 00:29:37.150 Deepika Sethi: So I’ve worked with Tabdue, I’ve worked with Power BI. I’ve not been working as a developer, but I have experience, firstly, creating basic reports, connecting, using connectors, and linking the tools to the databases, and fetching the data.
151 00:29:37.150 ⇒ 00:29:50.890 Deepika Sethi: And with Tableau, I can create some dashboards and sheets, you know, the quick ones, or I can also, you know, there’s one thing when you fetch the data, you would already know, like, you can use the SQL to kind of filter the data so that you really do not,
152 00:29:50.890 ⇒ 00:30:03.069 Deepika Sethi: dump everything into the tabloid, and your performance doesn’t degrade, so those things I have done in my career. Additionally, when it comes to reporting, there’s a specific tool called Exxiom, which is,
153 00:30:03.250 ⇒ 00:30:07.190 Deepika Sethi: a reporting tool, a regulatory reporting tool, I have worked with that.
154 00:30:07.660 ⇒ 00:30:11.090 Deepika Sethi: And, yeah, majorly, these three tools and.
155 00:30:11.090 ⇒ 00:30:15.500 Awaish Kumar: My question would be, like, what are you looking for in your new role?
156 00:30:16.530 ⇒ 00:30:36.250 Deepika Sethi: So, I’m looking for a com… you know, I was working in LSEC, and I realized that with the advent of latest technologies, I do feel that I had a need to really step back and understanding those, so when I look at my new role, I want to have some amalgation of my previous experience, so that I can utilize that. Plus.
157 00:30:36.340 ⇒ 00:30:55.949 Deepika Sethi: an enhancement with using new technologies like AI. It may not really be AI, some ML, some part of it, so that whatever I’m studying really doesn’t go waste, and, you know, I can use both of these and be at the forefront of technology when it comes to using the data.
158 00:30:57.480 ⇒ 00:31:01.839 Awaish Kumar: Okay, so… Are you recently working?
159 00:31:02.690 ⇒ 00:31:15.509 Deepika Sethi: So currently, I’m, pursuing my MS in Information Technology from Fordham University, so… and I’m on F1 visa in USA, and in May, I’ll be eligible for CPT.
160 00:31:17.450 ⇒ 00:31:21.209 Awaish Kumar: Okay, so right now, like, you will be working in a…
161 00:31:22.200 ⇒ 00:31:25.009 Awaish Kumar: Like, part-time selling, or how…
162 00:31:25.010 ⇒ 00:31:27.569 Deepika Sethi: Yeah, I can do the part-time thing currently.
163 00:31:28.820 ⇒ 00:31:31.490 Deepika Sethi: So, starting May, it can be full-time.
164 00:31:32.260 ⇒ 00:31:33.070 Awaish Kumar: Okay.
165 00:31:33.580 ⇒ 00:31:34.190 Deepika Sethi: Yeah.
166 00:31:34.900 ⇒ 00:31:36.339 Awaish Kumar: Yeah.
167 00:31:36.690 ⇒ 00:31:43.299 Awaish Kumar: And, okay, then I… I think…
168 00:31:43.900 ⇒ 00:31:46.210 Awaish Kumar: what I would like to know is that, like.
169 00:31:46.260 ⇒ 00:32:04.579 Awaish Kumar: to just clarify a few things. In this role, you, like, like, I think it’s already mentioned in the description, I don’t have to repeat myself, but, just to make things clear, it’s a hands-on job, so you have to execute things, although,
170 00:32:04.910 ⇒ 00:32:14.869 Awaish Kumar: Like, you have to talk to maybe stakeholders, maybe client, or internal team members to understand the requirements,
171 00:32:15.250 ⇒ 00:32:23.110 Awaish Kumar: But then, obviously, you have to go back and convert those requirements into technical specification, execute.
172 00:32:23.530 ⇒ 00:32:29.289 Awaish Kumar: Like, execution means you might be writing queries, you might be doing just some root cause.
173 00:32:29.510 ⇒ 00:32:33.980 Awaish Kumar: Analysis, or you will be maybe doing,
174 00:32:34.290 ⇒ 00:32:44.160 Awaish Kumar: some kind of, reporting in Google Sheets or any BI tool. So, like, this is what… make them…
175 00:32:44.550 ⇒ 00:32:49.829 Awaish Kumar: What it will look like if you are supposed to join the company.
176 00:32:50.890 ⇒ 00:33:03.280 Deepika Sethi: No, I think that makes sense. That’s what I was mentioning, right? That the reason I took a step back and being in a CADMI is to really understand those nitty-gritties, and not just the domain part of it, but also the technical side.
177 00:33:03.550 ⇒ 00:33:04.330 Awaish Kumar: Okay.
178 00:33:04.580 ⇒ 00:33:08.340 Awaish Kumar: I… so, I think I have everything.
179 00:33:09.630 ⇒ 00:33:13.480 Awaish Kumar: So, yeah, I will leave some time for you to ask any questions.
180 00:33:13.720 ⇒ 00:33:30.240 Deepika Sethi: Thank you for that. I do have a few questions. Actually, I think we talked a bit about data quality. One of my questions was, when you say that you are providing both data and AI services. Now, when it comes to data services, it’s still understandable, but when it comes to AI services, let’s say
181 00:33:30.320 ⇒ 00:33:35.979 Deepika Sethi: the user comes to you with an AI problem or something, but ultimately you realize that
182 00:33:36.150 ⇒ 00:33:45.249 Deepika Sethi: Or maybe their data is not really that good, so you don’t really get the results. How do you ensure that the client understands the data quality part of the work?
183 00:33:45.710 ⇒ 00:33:49.179 Deepika Sethi: Like, the data is… ultimately, it’s garbage in, garbage out, right?
184 00:33:50.030 ⇒ 00:33:56.890 Awaish Kumar: That’s the first, part of our… our discovery, right? Whenever we…
185 00:33:57.480 ⇒ 00:34:03.799 Awaish Kumar: we talk about, talk about a, any AI service or AI project.
186 00:34:04.030 ⇒ 00:34:09.479 Awaish Kumar: Right, like… Where the data will come from for this?
187 00:34:09.840 ⇒ 00:34:16.450 Awaish Kumar: project. Like, we need raw data, we need, business domain knowledge, we need,
188 00:34:16.800 ⇒ 00:34:20.429 Awaish Kumar: We need to define, all the different metrics, or…
189 00:34:20.659 ⇒ 00:34:26.579 Awaish Kumar: things we’re talking about. So that’s all part of discovery, like, we do that discovery with the client.
190 00:34:27.040 ⇒ 00:34:35.699 Awaish Kumar: with that discovery, they understand. If our definition is not correct, then obviously our tool is not going to work.
191 00:34:35.949 ⇒ 00:34:37.999 Awaish Kumar: If data isn’t cleaned.
192 00:34:38.620 ⇒ 00:34:54.530 Awaish Kumar: then it’s not going to work. So, whenever we do any AI project, obviously, we start with minimalist approach, like, you obviously take some kind of data, start working, show them
193 00:34:54.610 ⇒ 00:35:06.990 Awaish Kumar: some work, right? MVP, right? And then after that, obviously, you are going to propose, like, if we are to build a fully-fledged product out of it, it needs to be…
194 00:35:07.140 ⇒ 00:35:13.020 Awaish Kumar: We need to do this data engineering work as well, right? So, that’s… Hard goes.
195 00:35:13.020 ⇒ 00:35:31.120 Deepika Sethi: For example, like, let’s say the client is pretty much convinced that there’s… because I have faced as a client using Excel. There was obviously checks on the Excel, and they were manually doing a lot of work before putting it down directly into our database directly, so if client thinks that their data is pretty much good, but you.
196 00:35:31.120 ⇒ 00:35:37.779 Awaish Kumar: But maybe the process is not really that… Yeah, the only way… the only way to handle that is with FEX, right?
197 00:35:37.930 ⇒ 00:35:39.840 Awaish Kumar: Well, like, obviously this is…
198 00:35:40.110 ⇒ 00:35:50.569 Awaish Kumar: This is what may be part of disagreements, right? So, you could… you could have disagreement with your boss, your teammates, or the client, but this is how you are going to handle that.
199 00:35:51.250 ⇒ 00:35:55.000 Awaish Kumar: Right? It’s just with the numbers, right? You’ll go back.
200 00:35:55.690 ⇒ 00:36:02.030 Awaish Kumar: and have come up with some slides, or some Google Sheets, and they are going to show
201 00:36:02.220 ⇒ 00:36:08.119 Awaish Kumar: The difference is, where the mismatch is, where the gap is, and that’s how you convey your message.
202 00:36:08.590 ⇒ 00:36:15.189 Deepika Sethi: So, ultimately, if I get it right, even if it’s AI consultancy part of your job, you make sure that the data is first sorted out.
203 00:36:16.920 ⇒ 00:36:26.289 Awaish Kumar: Yeah, obviously, right? So, that’s an obvious part of it, as… as an AI, like, there are some projects where you don’t need that.
204 00:36:27.750 ⇒ 00:36:35.659 Awaish Kumar: like, that’s more, like, in ML, right? So normally, if somebody comes with, okay, I need some chatbot, or I need some…
205 00:36:36.400 ⇒ 00:36:42.879 Awaish Kumar: Something like, Which can… Handle some kind of…
206 00:36:43.820 ⇒ 00:36:58.810 Awaish Kumar: like, QA-based, question-answering-based something, some feature, that is pretty, like, doable, like, that’s… doesn’t involve, like, much of the data. I understand where you’re coming from, it’s more like the ML, like, if our data…
207 00:36:59.690 ⇒ 00:37:06.730 Awaish Kumar: is wrong, like, if it’s trained on the data, which is not really clean, and…
208 00:37:07.460 ⇒ 00:37:12.950 Awaish Kumar: Then, obviously, it’s… it’s not going to work as expected on the real data.
209 00:37:13.060 ⇒ 00:37:16.710 Awaish Kumar: So… Leah, but…
210 00:37:17.100 ⇒ 00:37:25.550 Awaish Kumar: Right now, we… we are looking to expand into ML projects, but right now, we are mostly focused on AI-related work.
211 00:37:26.360 ⇒ 00:37:26.950 Deepika Sethi: Oh.
212 00:37:27.360 ⇒ 00:37:38.569 Deepika Sethi: I think that was one thing. Another thing I had was, as you mentioned, you know, everyone is working at their own pace, in their own working style when it comes to Brain Forge.
213 00:37:38.630 ⇒ 00:37:50.530 Deepika Sethi: So, how does the growth at Braid Forge look like, since everyone is, like, working in a different country, in a different setting, and with the different expectations, I would say?
214 00:37:50.930 ⇒ 00:37:54.449 Awaish Kumar: The… at Brain Forge,
215 00:37:56.320 ⇒ 00:37:59.249 Awaish Kumar: if I say, like, how, like.
216 00:37:59.520 ⇒ 00:38:02.270 Awaish Kumar: when I say you will be working in your…
217 00:38:02.370 ⇒ 00:38:21.449 Awaish Kumar: on, like, as per your own, like, time zone or whatever, that does… that just means, like, we do collaborate, like, we have stand-ups, right? We do have client meetings. There are something which… which is, like, required, right? You can’t avoid that.
218 00:38:21.860 ⇒ 00:38:26.820 Awaish Kumar: You have to show up, the stand-ups, you have to, give updates,
219 00:38:26.990 ⇒ 00:38:34.710 Awaish Kumar: Even, like, doesn’t… like, you can give updates asynchronously, you don’t have to be in the meeting to give me updates.
220 00:38:35.180 ⇒ 00:38:47.819 Awaish Kumar: But things, like, the way it works is that, after the stand-up and after the client meetings, we want every team member to
221 00:38:48.180 ⇒ 00:39:00.019 Awaish Kumar: focus on the real work, instead of just being in meetings. So we try to avoid that, and try to come… so that’s why we try to communicate async. So, like, you don’t have to wait for me to, like.
222 00:39:00.250 ⇒ 00:39:02.309 Awaish Kumar: Take some time on my calendar.
223 00:39:02.440 ⇒ 00:39:07.149 Awaish Kumar: and then ask one question. You can write… write down if you want to create a…
224 00:39:07.310 ⇒ 00:39:14.330 Awaish Kumar: a short loomed for me, so you can send it, I can watch it at whatever time I wake up, and I can reply, right?
225 00:39:15.660 ⇒ 00:39:18.340 Awaish Kumar: So, these are the few things which we do.
226 00:39:18.930 ⇒ 00:39:24.319 Awaish Kumar: Obviously, Apart from that, the… the… In terms of growth.
227 00:39:25.690 ⇒ 00:39:28.389 Awaish Kumar: Like, we are pretty much, like,
228 00:39:28.680 ⇒ 00:39:37.260 Awaish Kumar: we have defined the carrier letters. It depends how you want to move forward. So, like, once… if you are in the company, then…
229 00:39:38.480 ⇒ 00:39:50.839 Awaish Kumar: if you want to take a leadership role, if you want to continue as an IC, if you want to… if you want to take a leadership role, how… that’s also defined, like, which way you want to go, what…
230 00:39:51.480 ⇒ 00:40:01.159 Awaish Kumar: What you choose to become, right? It’s all defined, and it depends on your interest, that… which path you want to take, and
231 00:40:01.730 ⇒ 00:40:03.350 Awaish Kumar: And yeah, I move forward.
232 00:40:04.600 ⇒ 00:40:08.720 Deepika Sethi: I just have one more question. Is it okay? Because we’re already on time.
233 00:40:10.230 ⇒ 00:40:16.319 Deepika Sethi: Okay, I’ll just be quick. Yeah, so when you talk about Brainforge AI, now, ultimately.
234 00:40:16.460 ⇒ 00:40:19.660 Deepika Sethi: You are dependent on the data from multiple
235 00:40:20.170 ⇒ 00:40:27.350 Deepika Sethi: databases that maybe, again, as we talked, Snowflake or Oracle, Now, Snowflake is pretty much…
236 00:40:27.480 ⇒ 00:40:29.460 Deepika Sethi: aggressive on doing that AI.
237 00:40:30.050 ⇒ 00:40:38.980 Deepika Sethi: how do you think those big companies, which already have data, are trying to move into AI impacts, brain force, you know, services.
238 00:40:39.810 ⇒ 00:40:41.490 Awaish Kumar: Norway, and, like.
239 00:40:42.180 ⇒ 00:40:50.440 Awaish Kumar: What they are doing is basically building off on top of whatever is already in the data warehouse, right?
240 00:40:51.090 ⇒ 00:40:54.269 Awaish Kumar: So, the pain hard work begins.
241 00:40:55.260 ⇒ 00:40:57.160 Awaish Kumar: Like, before that.
242 00:40:57.720 ⇒ 00:41:05.550 Awaish Kumar: like, the data… lending all of that data into Snowflake and modeling it.
243 00:41:05.830 ⇒ 00:41:12.629 Awaish Kumar: And converting into a usable… Meaningful format, right?
244 00:41:12.750 ⇒ 00:41:26.240 Awaish Kumar: is what is needed, right? They are building AI tools that can interact with your data, but as you mentioned, if the raw data is garbage, then your, basically, chatbot does nothing.
245 00:41:26.560 ⇒ 00:41:41.009 Awaish Kumar: Unless you have a data model data there. So, right. So that’s what we do. We come from a place where we do all the hard work, we bring in the data, so data for some people lives in an FTP server.
246 00:41:41.210 ⇒ 00:41:47.879 Awaish Kumar: Nobody’s connecting to that right now. Some of the data lives in, for example, in some…
247 00:41:48.100 ⇒ 00:41:55.680 Awaish Kumar: S3 routes from AS400, like, legacy database. Nobody has a connector for that. How will they…
248 00:41:55.980 ⇒ 00:41:59.369 Awaish Kumar: That’s how they are going to use that. So, like.
249 00:41:59.500 ⇒ 00:42:08.370 Awaish Kumar: you need somebody to move that data from ES400 to Snowflake, and then model it to make it use… to give it names.
250 00:42:08.640 ⇒ 00:42:16.809 Awaish Kumar: And the main… the last piece, the important thing is the context, like, the knowledge, the business domain knowledge.
251 00:42:17.040 ⇒ 00:42:24.309 Awaish Kumar: Snowflake does not have that, and they don’t have a way to capture it right now, right? Not any other company.
252 00:42:25.590 ⇒ 00:42:26.960 Awaish Kumar: brain
253 00:42:27.400 ⇒ 00:42:35.789 Awaish Kumar: obviously, for individual clients, that is different. We might have a generic knowledge, which everybody will have, but…
254 00:42:35.950 ⇒ 00:42:45.609 Awaish Kumar: When it comes to real domain knowledge, we do discoveries, we meet with people, we gather the requirements, we listen to their conversation, convert into
255 00:42:45.720 ⇒ 00:42:50.169 Awaish Kumar: Some useful documentation, and then we…
256 00:42:50.680 ⇒ 00:42:55.290 Awaish Kumar: Use that alongside the data, which is modal and clean.
257 00:42:55.510 ⇒ 00:42:59.790 Awaish Kumar: And then mix up both makes it, like, really useful for the AI.
258 00:43:00.060 ⇒ 00:43:03.189 Awaish Kumar: Then we provide AI services, obviously. So…
259 00:43:03.760 ⇒ 00:43:17.799 Awaish Kumar: Obviously, we are… maybe we… we have AI team, they are… maybe they use… they can build their own systems, but obviously, they can just plug in to Snowflake. Like, they can use existing tools as well, like…
260 00:43:17.920 ⇒ 00:43:20.699 Awaish Kumar: That is the hard work which we did. We…
261 00:43:20.830 ⇒ 00:43:29.519 Awaish Kumar: modeled the data, we brought in the information… the knowledge, right? We defined it in a way that is… that is, like.
262 00:43:30.430 ⇒ 00:43:35.409 Awaish Kumar: like, what you say, that’s… that’s understandable by the AI agents.
263 00:43:35.930 ⇒ 00:43:37.370 Awaish Kumar: That’s what was needed.
264 00:43:37.540 ⇒ 00:43:44.400 Awaish Kumar: The final layer, it can be our front end developer, or it can be someone else’s front-end developer.
265 00:43:45.430 ⇒ 00:43:46.480 Deepika Sethi: admit center.
266 00:43:46.670 ⇒ 00:43:49.339 Deepika Sethi: I think that was all I had.
267 00:43:49.340 ⇒ 00:43:54.490 Awaish Kumar: Okay, so I will leave my feedback,
268 00:43:54.950 ⇒ 00:44:02.420 Awaish Kumar: to the… the operational… operations team, and they are, like, the Rico from our team will… will come back with the next steps.
269 00:44:03.510 ⇒ 00:44:04.200 Deepika Sethi: Thank you.
270 00:44:04.880 ⇒ 00:44:05.390 Awaish Kumar: Right.