Meeting Title: Brainforge Interview w- Sam Date: 2026-03-04 Meeting participants: Ned, Samuel Roberts
WEBVTT
1 00:00:11.940 ⇒ 00:00:12.710 Samuel Roberts: Hello?
2 00:00:12.710 ⇒ 00:00:13.230 Ned: Hey.
3 00:00:15.800 ⇒ 00:00:17.170 Ned: Hey! Hey, Simon.
4 00:00:17.170 ⇒ 00:00:18.169 Samuel Roberts: How are you?
5 00:00:18.570 ⇒ 00:00:20.760 Ned: I’m doing great, thank you for asking. How are you?
6 00:00:21.000 ⇒ 00:00:24.060 Samuel Roberts: Good, good. Sorry, one sec, my connection seems a little…
7 00:00:24.200 ⇒ 00:00:27.719 Samuel Roberts: My whole computer’s a little slow right now, I think, more than just me, but…
8 00:00:27.960 ⇒ 00:00:31.940 Samuel Roberts: There we go, okay. Now I see and hear you better.
9 00:00:32.870 ⇒ 00:00:33.560 Ned: Perfect.
10 00:00:34.900 ⇒ 00:00:35.820 Ned: Great.
11 00:00:35.820 ⇒ 00:00:38.719 Samuel Roberts: So, yes, thank you for taking the time,
12 00:00:38.980 ⇒ 00:00:43.889 Samuel Roberts: My name is Sam Roberts, I’m the AI tech lead here at Brainforge.
13 00:00:45.610 ⇒ 00:00:55.850 Samuel Roberts: Yeah, I mean, I can jump in a little bit more, tell you more about stuff, but I figured let’s just do some quick introductions, so why don’t you give me a little bit of your background, just a little bit of an intro.
14 00:00:57.130 ⇒ 00:01:10.580 Ned: Amazing. Definitely, yeah, I can definitely provide you my background. So, Sam, I have around, like, you know, 8 years of experience building and focusing on, like, you know, multiple, you know, end-to-end AI products itself and solutions as well.
15 00:01:10.580 ⇒ 00:01:14.730 Ned: That actually solves the actual business problems. That’s what I have looking forward to.
16 00:01:14.730 ⇒ 00:01:37.409 Ned: Also, like, you know, currently as the lead data scientist at Wound AI, so my focus has been on, like, you know, high-scale automations and operational efficiencies. I have built robust ETL pipelines all by myself using Apache, Spark, and AWS Kinesis, and deployed these supervised, you know, models like XTBoost and achieved 91% accuracy, directly cutting the malware intervention by 40%.
17 00:01:37.730 ⇒ 00:01:55.010 Ned: Prior to that, so at Pentax Solutions, I spearheaded the, like, you know, the NLP action tools for sentiment analysis and, you know, document matching as well. I also spent time at Visa, where I built the neural networks, you know, for call routing and language-to-SQL transition tools as well.
18 00:01:55.010 ⇒ 00:02:03.459 Ned: help non-technical users query the database all by themselves. They don’t have to come back to me, they can just utilize that particular, you know, project that I built.
19 00:02:03.480 ⇒ 00:02:17.199 Ned: Regarding the specific requirements at, like, you know, for Brain Forge, yes, I have read the video very well as well. I’m the first and foremost, like, you know, what I put, like, you know, call myself a consultant, because that’s what I have done so far.
20 00:02:18.150 ⇒ 00:02:29.229 Ned: while my, like, you know, core is based in Python and ML frameworks like PyTorch and TensorFlow, I’m definitely a full-stack engineer, because I like to take the ownership from the start till very end.
21 00:02:29.380 ⇒ 00:02:35.930 Ned: So, I have both the front ends as well, utilizing TypeScript, ReactJS, and Node.js.
22 00:02:36.140 ⇒ 00:02:48.740 Ned: Also, I’m deeply immersed on the LLM space lately, working with, like, Langchain, OpenAI, and vector stores like Fires, MiddleWest, Pinecone, as well, for a rag-based system. I have.
23 00:02:48.740 ⇒ 00:02:49.470 Samuel Roberts: Great.
24 00:02:49.470 ⇒ 00:03:03.970 Ned: developed multimodal agents as well, specifically, like, you know, one using with the stable diffusion and the lava as well, where I had, like, you know, the custom evaluation loops to ensure the output was quality.
25 00:03:04.200 ⇒ 00:03:24.870 Ned: So, again, like, you know, with respect to startup phase itself. So, I have been within, like, you know, Pentec Solutions, then OneDI, which were actually startups. So, I started out with them, well, they had, like, around… at Pentec, they had 7 people at that time, when I joined there, and OneDI, they had just 5 people, initially, when I joined there.
26 00:03:24.970 ⇒ 00:03:38.879 Ned: Yeah. So, speaking about myself, Pentec and Wound AI both have provided me a very, very good, like, consulting sense. I have become, like, the face of the organization at Wound AI. That’s what I was doing at Pentec Solutions as well.
27 00:03:38.880 ⇒ 00:03:46.830 Ned: So, speaking about Wound AI and Fintech Solutions, those are basically service providers which actually have multiple different clients with multiple different areas.
28 00:03:46.900 ⇒ 00:04:04.700 Ned: So, multiple people come back with their particular use case. It’s our job to, you know, solutionize those use cases. So, you know, being the face, I was actually going into meetings with the, you know, end clients to discover what sort of problem they are having at the moment, so that I can pick up the pain points, and right afterwards provide them the solution.
29 00:04:04.700 ⇒ 00:04:15.489 Ned: what I normally do, so, with respect to the solution itself, I know that the end clients, they love to have some sort of, you know, visualizations, or hands-on experience as well, you know that.
30 00:04:15.490 ⇒ 00:04:18.269 Ned: So, what I normally do, I just take, like, you know.
31 00:04:18.269 ⇒ 00:04:30.210 Ned: maybe, like, 2 hours, 3 hours, maybe 6 hours, maybe 1 day, 1 and a half day, based on the use case, to provide some sort of POC. Not the POC, because, you know, POCs are not scalable at all.
32 00:04:30.270 ⇒ 00:04:41.269 Ned: I’d like to move forward with MVPs, so that those can be scalable, utilizing the tools that we actually need to utilize, like, in the longer end as well. So I’d like to provide them something.
33 00:04:41.450 ⇒ 00:04:54.160 Ned: And right afterwards, gather their feedbacks and their approvals. Once approved, we can definitely move towards the production-grade systems. So I have moved, like, I have worked on, you know, production-grade systems all by myself, hands-on, as well.
34 00:04:54.300 ⇒ 00:05:04.209 Ned: Also, like, speaking about myself, I’m a very high-paced person, so I love to work, like, you know, quite a lot, let’s say, starting from 8 AM in the morning till 8pm in the night.
35 00:05:04.210 ⇒ 00:05:19.019 Ned: So that I can deliver everything. I like to take ownership. Let’s say my manager has come back and said that, Ned, you need to provide the implementation in 3 days. Let’s say it doesn’t seem like, you know, quite possible as well, but
36 00:05:19.350 ⇒ 00:05:43.199 Ned: I know what tools to be used, what techniques to be used to ship that particular, you know, MVP out in 3 days, so that I can just get done with this. Also, like, speaking about myself, I’m that person who doesn’t take any leaves as well. Not in the Christmas period as well, because I think that my mentality is that, you know, there should be someone within the team who is actually taking care of the business while everybody’s on leaves.
37 00:05:43.310 ⇒ 00:05:47.920 Ned: And that’s the one. Okay, okay, cool, cool.
38 00:05:48.470 ⇒ 00:05:49.120 Samuel Roberts: Yeah.
39 00:05:49.490 ⇒ 00:06:06.840 Samuel Roberts: Thank you, yeah, that’s very helpful. A lot of context there that I can now wrap my head around. I guess let’s dig in a little bit. So you mentioned a few things you built, but I guess what have you spent, like, where in the AI stack have you spent the most time building versus just kind of experimenting with?
40 00:06:07.500 ⇒ 00:06:26.170 Ned: So, with respect to, like, experimentations itself, I have been experimenting quite a lot with respect to new tech stacks, new tools that are coming up in the market. I have always been, like, you know, eyeing out for any sort of tools or techniques that are coming up from, let’s say, Azure GCP or AWS itself, to make sure that I definitely understand those new techniques.
41 00:06:26.170 ⇒ 00:06:40.770 Ned: And later, afterwards, I can just supply those within the projects itself, or the product that I’m currently building at the moment. So that’s my mentality. I’m always in for R&Ds, so that I can research and make myself, you know, up-to-date. My goal is to become a director in 5 years’ time.
42 00:06:40.770 ⇒ 00:06:59.780 Ned: So, I have been a lead in Warned AI. So how can I become a director if I do not know about the latest tech tags? So I need to make updated as well. So, if I’m leading a team, I need to provide them, you know, good enough mentorship with respect to the new and the techniques that are basically, like, just rolled out within the market, and what are the, you know, benefits with them?
43 00:06:59.840 ⇒ 00:07:06.069 Ned: Speaking about, you know, what I have been, you know, doing quietly, with respect to the projects itself.
44 00:07:06.570 ⇒ 00:07:24.599 Ned: I have been, you know, utilizing the… or maybe, like, you know, developing the pipelines all by myself, EDL pipeline, maybe with the help of, like, you know, Apache Spark, and also with the help of Airflow as well. That’s what I have been doing. I have utilized PySpark as well, if, you know, if we are moving towards Azure itself.
45 00:07:24.820 ⇒ 00:07:38.950 Ned: although I have been working within the, you know, core NLP, GenAI, RAGs, also within, you know, Agent AKI as well. I can definitely explain to you the project that I have done recently. That would clear up, like, you know, a lot of things.
46 00:07:39.480 ⇒ 00:07:40.160 Ned: Okay.
47 00:07:40.160 ⇒ 00:07:51.090 Samuel Roberts: Let’s… yeah, let’s keep going. I have some of… I have a bunch of questions I need… I want to get through. I guess… so, like, yeah, experience seems… seems great. I guess, let’s talk about,
48 00:07:51.130 ⇒ 00:08:05.419 Samuel Roberts: non-technical people, and interacting with stakeholders that, you may have to explain the limitations of the technology that they’re hearing all about all the time. How do you go about, explaining to non-technical stakeholders what that,
49 00:08:05.750 ⇒ 00:08:07.930 Samuel Roberts: The limitations, yeah.
50 00:08:08.420 ⇒ 00:08:19.750 Ned: I have been working within consultancies for quite a long time at 1DI and Protect Solutions, so I have met, like, you know, lots and lots of non-technical people, more non-technical than technical people. Sure.
51 00:08:19.750 ⇒ 00:08:34.190 Ned: Yeah, definitely, because CEOs can come up and say that, you know, we have heard about AI, it’s very amazing, looks very amazing, can you please, like, do something with the help of AI and solve our problem? And I’m like, yeah, we can definitely do this, but according to our process, we need to run through with respect to the solution architecture as well.
52 00:08:34.190 ⇒ 00:08:39.040 Ned: So that we can make them aligned. Because they’re the ones who are actually paying for it. So let’s say AWS…
53 00:08:39.400 ⇒ 00:08:45.959 Ned: less amounts, Azure, many, many, like, much more hefty amounts, because Azure is very expensive.
54 00:08:45.960 ⇒ 00:09:10.919 Ned: So, with respect to architectures or solutions, if I need to put it in front of the, you know, non-technical clients. So, what I normally do with non-technical stakeholders, I go with them, like, you know, with some sort of a plan. Non-technical users, they do not understand technologies, but they understand benefits, they understand ROIs, so I explicitly list out the ROIs and benefits with respect to the architecture we are pitching them, or the solution
55 00:09:10.920 ⇒ 00:09:11.870 Ned: you’re pitching them.
56 00:09:11.950 ⇒ 00:09:25.890 Samuel Roberts: So that they can understand what sort of impact it would provide it to them, what’s the cost, what’s, like, you know, how, moving forward, how much of their efforts will be reduced with the help of a particular automation that is, you know, built with respect to, like, let’s say, Agent AKI.
57 00:09:25.970 ⇒ 00:09:40.909 Ned: So, with respect to my diagrams, I tend to use, like, you know, many diagrams as well. So what I normally do, I just, like, you know, build up these solution architecture diagrams, utilizing just simple words. Let’s say they don’t have to understand what a, you know, database is.
58 00:09:40.910 ⇒ 00:09:48.669 Ned: It’s just like, you know, a storage box for them, where you just put all the data, and right afterwards, you’re just storing up the data with, you know, lots and lots of stacks within there.
59 00:09:48.690 ⇒ 00:10:00.649 Ned: If there’s any, let’s say, like, you know, within Agentech AI, we know that we have agents, and agents are basically the brains, so I can just, like, you know, list down as brains that they are thinking, whatever you curry, whatever you push down.
60 00:10:00.650 ⇒ 00:10:19.269 Ned: it would straight away go to the agent, who is actually acting as the brain, so that he can understand what sort of, you know, agentic nodes he wants to go forward with. And right afterwards, he’s going to, like, you know, retrieve the data, push it across to LLM. LLM would analyze the data, and right afterwards, it would present you the analysis in a nicer format.
61 00:10:19.340 ⇒ 00:10:24.749 Ned: So that’s how I basically, like, you know, communicate with them, so that they can understand in their easy languages.
62 00:10:25.260 ⇒ 00:10:25.840 Samuel Roberts: Great.
63 00:10:27.600 ⇒ 00:10:38.100 Samuel Roberts: Great, that’s good. Let’s talk about, there’s lots of new stuff coming out all the time in this industry. Are there… is there an example of some kind of, like.
64 00:10:38.150 ⇒ 00:10:48.280 Samuel Roberts: Either technology, or a tool, or framework, or trend in general that, you might have been excited about, but decided not to adopt for certain reasons in a project.
65 00:10:50.070 ⇒ 00:10:51.000 Ned: Ugh…
66 00:10:51.010 ⇒ 00:10:56.560 Ned: That’s a good one, I must say. So definitely, like, you know, I have been seeing lots and lots of advancements within
67 00:10:56.560 ⇒ 00:11:19.460 Ned: AI, so there are lots and lots of, like, you know, newer models are coming up. So, let’s say if we talk about, like, you know, newer models that are coming up in the market, I would definitely not use them, like, all of a sudden. So, firstly, what I would do, I would, like, you know, consider them, definitely, with respect to my own system, you know, have some sort of testing on the top of it, and right afterwards, I’m going to make a decision whether if I’m going to go forward with this or not.
68 00:11:19.710 ⇒ 00:11:20.850 Ned: So…
69 00:11:20.930 ⇒ 00:11:30.600 Ned: like, you know, I have a couple of examples as well, like, you know, which I have… which have been coming up in my mind with respect to my, you know, experiences as well, and I would definitely love to
70 00:11:30.600 ⇒ 00:11:47.799 Ned: showcase them. So, a good example that is coming up in my mind is, like, you know, auto-GBD-style autonomous agents that are coming up in the market. I was excited about it because, like, you know, it showed how you could chain the LLMs with tools and memory to pursue the goal autonomously.
71 00:11:47.800 ⇒ 00:11:54.800 Ned: Conceptually, it felt like, you know, the future of agentic systems, but I didn’t adapt them. I have my own reasons.
72 00:11:54.830 ⇒ 00:12:05.089 Ned: The very first one would be, like, you know, unreliable control that you can get. So, it could loop or drift from the objective. That’s what I have seen, during my own testing.
73 00:12:05.400 ⇒ 00:12:21.050 Ned: it is very hard to evaluate as well, because there’s no clear way to measure the success or failure deterministically. And yes, the cost and latency as well. So too many LLM calls per task for, like, let’s say, you know, very small business cases as well.
74 00:12:21.050 ⇒ 00:12:28.249 Ned: So that’s why I actually, like, dropped it. But if there’s any sort of advancement within that, I would love to, you know, try that again.
75 00:12:28.710 ⇒ 00:12:29.990 Samuel Roberts: Sure, great.
76 00:12:30.270 ⇒ 00:12:35.420 Samuel Roberts: Okay, well, actually, building on that then, how do you…
77 00:12:36.330 ⇒ 00:12:44.479 Samuel Roberts: when do you decide something is production ready? A tool or something? So this, like I said, things are coming out quickly, so, you know, you mentioned,
78 00:12:44.900 ⇒ 00:12:45.860 Samuel Roberts: what, what…
79 00:12:46.890 ⇒ 00:12:58.140 Samuel Roberts: what is the kind of basic criteria? You don’t have to get, like, too specific nitty-gritty, but, you know, you kind of mentioned you look at it and do that, but, like, what might you have to evaluate, at least a high level?
80 00:12:59.340 ⇒ 00:13:00.270 Ned: Sure, sure.
81 00:13:00.310 ⇒ 00:13:19.030 Ned: So, Sam, I definitely have a very, like, you know, stakeholder or client-centric approach, because whatever work that I’m currently doing, it’s for the benefits for them. I’m getting paid, I know that, but still, like, it’s for the sake of that particular client itself, or internal stakeholder, if I’m building up any sort of product itself, or any project internally.
82 00:13:19.030 ⇒ 00:13:27.659 Ned: But still, like, you know, there are certain stakeholders against which, like, they have lots and lots of hopes, so I need to manage those hopes as well. So what I normally do…
83 00:13:27.660 ⇒ 00:13:35.650 Ned: like, my style of working. So, what I normally do? So, at, you know, Pentec Solutions, as well as 1DI, because I have, like, taken
84 00:13:35.650 ⇒ 00:13:44.339 Ned: key takeaways from PinTech against the bottlenecks that we used to have, and right afterwards, I applied those solutions on one day to see, like, you know, how much we can progress.
85 00:13:44.340 ⇒ 00:13:55.369 Ned: So what I did, actually, I made this a particular policy that we should have four environments. And this is working for us, definitely. Four environments, why? The very first one could be, like, the dev environment.
86 00:13:55.370 ⇒ 00:14:07.899 Ned: So this would be, like, you know, if I’m deploying anything with respect to the, you know, project itself, I would love to see the impact of it within the dev environment, to see, like, you know, whenever it is going to be pushed to the production environment, how would it react?
87 00:14:08.130 ⇒ 00:14:28.079 Ned: So, I love to do testings on dev environment. So, let’s say, like, you know, if I’m very happy on my testings, and if I see that, you know, all of this implementation that I have done so far, it’s, you know, up to the mark, I’m going to push it across to the test environment, where the internal QAs can jump in, and they can test it across. If they have any feedbacks, I would love to have it.
88 00:14:28.080 ⇒ 00:14:30.700 Ned: So that I can, you know, go back, straight away, implement that.
89 00:14:30.830 ⇒ 00:14:39.699 Ned: Right forward, like, you know, if they have tested everything, if they’re happy with their sign-off, I can push it across to the UAT environment. So what UAT environment is?
90 00:14:39.700 ⇒ 00:15:01.710 Ned: It is actually for the stakeholders who are going to use the system in production environment. So I would definitely love to validate each and everything with respect to any sort of edge case scenarios as well, in front of them, so that they can have the taste of everything at first, they can gain the confidence, and during that phase, I can provide them the, you know, trainings as well, how to use the system. Even, like, not within the UAT phase.
91 00:15:01.790 ⇒ 00:15:09.080 Ned: That’s what I normally do. I’m always very, like, you know, advocate of providing the demos. So during the development as well.
92 00:15:09.290 ⇒ 00:15:15.070 Ned: even if on the local environment, I definitely provide the demos to the stakeholders weekly.
93 00:15:15.250 ⇒ 00:15:19.349 Ned: So, going in an agile manner to see, like, if there’s any feedback.
94 00:15:19.390 ⇒ 00:15:34.479 Ned: If yes, I can cater that. If no, we are going towards the right direction. So they’re going to, like, set up the directions, because they will be the end users of that particular system, and they will gain the confidence that we are going in the right direction. Second thing, they will be getting the rightly updates as well.
95 00:15:34.480 ⇒ 00:15:46.259 Ned: So, I always, like, you know, provide them the demos weekly. Like, I like to call them show and tells. Demos are, like, you know, the system is ready. Show and tells just means, like, you know, I can show you something, you can tell me if it’s…
96 00:15:46.260 ⇒ 00:15:47.169 Ned: good or not.
97 00:15:47.350 ⇒ 00:15:48.219 Ned: So, right now.
98 00:15:48.220 ⇒ 00:15:48.890 Samuel Roberts: Yeah.
99 00:15:49.440 ⇒ 00:16:01.249 Ned: Yeah, it always helps us, definitely going towards the administrative parts. So, within the sprints, if they have any sort of, like, you know, feedbacks and stuff, we can cater those in the sprints. If not, we can just push it across.
100 00:16:01.250 ⇒ 00:16:14.139 Ned: So that we can manage all the expectations. And I love to, you know, have the discussions with them to let them know that, you know, this sprint will cater this, this sprint will cater this. Eventually, we’ll ship out the system in the, you know, agreed timelines, definite.
101 00:16:14.510 ⇒ 00:16:26.899 Ned: So right afterwards, let’s say UAT environment, everyone is happy. Right afterwards, this is the metric that, you know, I would take, and, you know, whenever the approvals come in, I’ll definitely deploy this to the production environment.
102 00:16:27.120 ⇒ 00:16:27.660 Samuel Roberts: Cool.
103 00:16:27.830 ⇒ 00:16:28.990 Samuel Roberts: metric.
104 00:16:29.490 ⇒ 00:16:39.859 Samuel Roberts: Thank you, yeah, excellent. Okay, so we’re about halfway, so I kind of want to switch and see if you have any questions that I can answer about Brainforge, about the role, anything like that.
105 00:16:40.250 ⇒ 00:17:00.350 Ned: Yeah, definitely, definitely. So I would love to know about the team more, like, what sort of what you have currently. I have done some, like, you know, research on Brainforge as well. I know that you’re in a startup phase at the moment. That’s what I loved about, you know, Brainforge, because I wanted to jump in into a startup. You have, like, around 20 people within the team itself.
106 00:17:00.350 ⇒ 00:17:12.599 Ned: At the moment. So, you want the other person to take end-to-end ownership. That’s what I actually loved about, you know, Brainforge as well. But I still would love to know about, like, you know, what is the stream structure, what is the hierarchy?
107 00:17:13.079 ⇒ 00:17:14.969 Samuel Roberts: Yeah, yeah, so we’re,
108 00:17:15.519 ⇒ 00:17:21.709 Samuel Roberts: Yeah, so we’re relatively small, as you mentioned. We’re kind of broken engineering-wise into two teams. So there’s the data team.
109 00:17:21.899 ⇒ 00:17:38.139 Samuel Roberts: And they do data pipelines, ETL, there’s some analysis, but that’s sort of a separate team. But for the AI side, AI automation team, basically, there’s only a handful of us. It kind of spun out of the, kind of.
110 00:17:38.309 ⇒ 00:17:48.929 Samuel Roberts: things we were doing internally that were helping, kind of… we started as a data consultancy initially, so, that kind of became, okay, well, we’re building these things, we need people to do it, and then as we had more people.
111 00:17:48.929 ⇒ 00:18:02.189 Samuel Roberts: clients started asking, so we started building and kind of selling these services. And so, at any given time, there’s a few different AI clients, and we’re finding more and more as we grow. Perfect. But we’re looking at all kinds of
112 00:18:02.269 ⇒ 00:18:18.729 Samuel Roberts: you know, implementations of AI tools and automation. So, for example, we’ve done some where it’s basically, you know, someone will come and say, like, I know I can use Claude, I do this, I have these prompts, and we’ve done some work with basically just automating that flow for them, so they don’t have to
113 00:18:18.779 ⇒ 00:18:28.439 Samuel Roberts: Copy, paste, copy, paste. We’ve done other ones that are more, rag-heavy chatbots over, you know, tons of documents and information for.
114 00:18:28.439 ⇒ 00:18:43.479 Samuel Roberts: customer service agents, so, you know, you’re talking to someone on the phone and they need to look something up, that’s, that’s some stuff we’ve done there. We’ve done some others where it’s more, MCP-driven, pulling in data, including some, you know, data pipelines and stuff that way, being able to chat over the data and all that sort of stuff.
115 00:18:43.479 ⇒ 00:18:54.979 Samuel Roberts: So we’re really pretty flexible, it’s kind of, you know, what people want. We’re, we’re, you know, it’s a new… lots of things are changing so quickly, so we’re learning things and applying them pretty quickly.
116 00:18:56.519 ⇒ 00:19:00.839 Samuel Roberts: Yeah, I think… does that answer most of your question there, or is there… did I miss anything?
117 00:19:01.380 ⇒ 00:19:09.960 Ned: Yeah, most of it, definitely. I love the consulting part, I love the, like, you know, having multiple use cases, multiple different solutions for any sort of, like, end clients.
118 00:19:09.990 ⇒ 00:19:28.569 Ned: That’s the environment that I’m looking forward to have, because, like, I’ve been in this particular environment for quite a long time. I would definitely love to be in the same environment where I’m consulting, looking at multiple different use cases, and right afterwards, providing the implementations for the end users, and having, like, you know, them smiling back to us.
119 00:19:28.570 ⇒ 00:19:32.459 Ned: Stating that this particular, like, you know, solution has worked very well for us.
120 00:19:32.460 ⇒ 00:19:36.880 Samuel Roberts: Sure, yeah, definitely, definitely. Great, yeah. Other, other questions, or…
121 00:19:37.900 ⇒ 00:19:47.960 Ned: With respect to the expectations of this role, so let’s say, like, you know, 3 months down the line, what sort of expectations do you have with the role that I have applied for for the next 90 days?
122 00:19:48.590 ⇒ 00:20:02.109 Samuel Roberts: Yeah, so, there’s usually some amount of, like a testing phase, probationary period sort of thing, but really, it’s no different than it would be if it was, you know, post that. You know, you’d be…
123 00:20:02.450 ⇒ 00:20:17.069 Samuel Roberts: onboarded, and obviously we’re, like you said, a startup, so we’re still kind of figuring some of that stuff out, so onboarding is different than when I joined over the summer and things like that, but we’re starting to structure that a little bit more for people, so hopefully, you know, you’d get set up with everything. We have,
124 00:20:17.090 ⇒ 00:20:29.509 Samuel Roberts: you know, we use basic tools, Slack, GitHub, obviously, like, all these things you need to get onboarded to. And then you’d probably jump right into some client work. We do kind of have a split between, sort of, the internal tooling.
125 00:20:29.710 ⇒ 00:20:32.059 Samuel Roberts: That we build, and the client…
126 00:20:32.200 ⇒ 00:20:34.220 Samuel Roberts: Work that we do, so a lot of us.
127 00:20:34.220 ⇒ 00:20:34.880 Ned: Yep.
128 00:20:34.880 ⇒ 00:20:46.809 Samuel Roberts: on the AI team kind of straddle both sides. Sometimes we’re doing work for the client, or a client, sometimes we’re doing work for multiple clients, and the internal team, you know, so it’s bouncing around a lot. And I think, you know.
129 00:20:47.020 ⇒ 00:20:56.919 Samuel Roberts: presuming, you know, in the first 90 days, things go well that way, you know, we… we tend to load a lot of autonomy onto people, so, you know, as, like, the…
130 00:20:57.300 ⇒ 00:21:16.569 Samuel Roberts: the tech lead, I’m not necessarily being like, you know, Code Monkey, go do this. It’s more like, okay, we’ll talk through a plan, we can evaluate the plan, think about that sort of stuff. You know, we’re using a lot of coding agents and experimenting with different things that way, so it’s a lot of, okay, this is an easy one I can feed to cursor and just let it run. And we’re trying to…
131 00:21:16.740 ⇒ 00:21:29.620 Samuel Roberts: as part of the internal tooling, figure out some of those processes. So you’d be involved in that. You know, the client work changes as clients come and go, so it’s hard to kind of pin that down, but, I imagine it’s, you know.
132 00:21:29.740 ⇒ 00:21:34.940 Samuel Roberts: And I don’t see you being, like, doing the same task over and over again, if that makes sense. It’d be lots of different stuff, so…
133 00:21:34.940 ⇒ 00:21:36.270 Ned: Understandable, yes.
134 00:21:36.870 ⇒ 00:21:46.770 Ned: I must say this, like, you know, you’re a tech lead, I know that, and you’re one of the most humble tech leads that I have, like, you know, met with recently.
135 00:21:46.810 ⇒ 00:21:55.710 Samuel Roberts: That’s interesting, thank you for that. I come from a more, like, product startup background, so I’m used to being, you know.
136 00:21:55.880 ⇒ 00:22:04.800 Samuel Roberts: like, the CTO of a two-person company. And so it’s very… that’s a very humbling experience in a lot of ways, because you’re doing a ton, and it’s not fast enough, and you gotta keep going faster. And so…
137 00:22:04.800 ⇒ 00:22:05.180 Ned: Yep.
138 00:22:05.180 ⇒ 00:22:22.329 Samuel Roberts: coming to Brainforge is, you know, more of a consultancy, and I don’t feel like I’ve… you know, I didn’t rise through the ranks of something else, if that makes sense. I don’t have that kind of ego, and I think that’s a big thing of Brainforge in general, like, there’s not a lot of egos, you know, you don’t want to deal with that, I feel like, so…
139 00:22:22.330 ⇒ 00:22:25.159 Ned: I love the culture already for brain fries.
140 00:22:25.160 ⇒ 00:22:35.450 Samuel Roberts: Yeah, no, I definitely appreciate it here. You know, I’ve started my own companies, and I know that’s not something that happens easily. It’s intentional, and you gotta work at it, even when it’s small.
141 00:22:35.580 ⇒ 00:22:38.240 Samuel Roberts: And I think we have a pretty good one here, so, yeah.
142 00:22:38.760 ⇒ 00:22:40.030 Ned: Amazing, amazing.
143 00:22:41.310 ⇒ 00:22:42.230 Ned: Right?
144 00:22:42.320 ⇒ 00:22:47.450 Ned: Just to let you know, so my contract will be ended with the one AI by the end of next week.
145 00:22:47.480 ⇒ 00:23:02.989 Ned: So, I’m more than happy to, you know, pick up anything from the week after. I’ll definitely look forward to have, like, you know, the other runs. I know that it was within the JD itself that there would be another, like, you know, role-focused technical interview, and then another panel interview. I would definitely…
146 00:23:03.590 ⇒ 00:23:04.810 Ned: Soon as possible.
147 00:23:05.020 ⇒ 00:23:14.999 Samuel Roberts: Yeah, yeah, we try to keep them moving as quick as possible. The real… the only bottleneck is just scheduling time, you know, synchronous time like this. But we’re… we’re trying to move things along relatively quickly.
148 00:23:15.120 ⇒ 00:23:25.370 Samuel Roberts: You know, we don’t like to drag it out like some companies do. So yeah, that, I mean, you nailed it. The last thing I was gonna say was the process, so you’ve already got that down, that’s good.
149 00:23:25.700 ⇒ 00:23:41.470 Samuel Roberts: So yeah, so presuming I bring everything back and it goes well, yeah, you’d get… you’d, they’d… someone from the recruiting team would reach out to you, scheduling that next one, with another member of the team, and then after that would be, yeah, the panel, which I believe I’d be on, so,
150 00:23:41.470 ⇒ 00:23:47.400 Samuel Roberts: And there’s probably some sort of technical challenge in there, I forget exactly where it falls, but yeah, you know, I don’t…
151 00:23:47.400 ⇒ 00:23:51.379 Samuel Roberts: Just something we can more talk about during an interview than anything else, you know?
152 00:23:51.380 ⇒ 00:23:52.150 Ned: So, nope.
153 00:23:52.380 ⇒ 00:23:56.850 Samuel Roberts: Yeah, I think that pretty much covers everything I had. Do you have any other…
154 00:23:57.820 ⇒ 00:24:08.200 Ned: No, I think, like, you know, pretty much covered, you know, as our discussions has gone. So you have, like, explained basically everything very beautifully. So I’m looking forward to have the next part of that ASAP.
155 00:24:08.490 ⇒ 00:24:10.820 Samuel Roberts: Okay, yeah, hopefully, hopefully it goes well, yeah.
156 00:24:10.940 ⇒ 00:24:11.660 Samuel Roberts: Alright.
157 00:24:11.950 ⇒ 00:24:13.189 Samuel Roberts: Thank you so much!
158 00:24:13.520 ⇒ 00:24:14.869 Ned: Thanks very much. Have a good day.
159 00:24:14.870 ⇒ 00:24:16.740 Samuel Roberts: Have a good one. Yep, you too. Bye.
160 00:24:16.740 ⇒ 00:24:17.290 Ned: Bye.