Meeting Title: BF Technical Screen Date: 2025-10-30 Meeting participants: Samuel Roberts, Awaish Kumar, Rico Rejoso


WEBVTT

1 00:00:18.550 00:00:20.010 Awaish Kumar: Hello?

2 00:00:21.220 00:00:22.140 Samuel Roberts: Hey!

3 00:00:23.460 00:00:24.410 Samuel Roberts: How you doing today?

4 00:00:24.410 00:00:26.949 Awaish Kumar: I’m good, how about you?

5 00:00:27.400 00:00:28.540 Samuel Roberts: Doing alright.

6 00:00:30.940 00:00:33.140 Samuel Roberts: Yeah Waiting on the room.

7 00:00:33.140 00:00:36.030 Awaish Kumar: So, like… We can discuss, like.

8 00:00:36.030 00:00:36.560 Samuel Roberts: Yeah, totally.

9 00:00:36.560 00:00:44.450 Awaish Kumar: It’s basically, like… We have to decide on what AI challenge should we…

10 00:00:44.870 00:00:50.550 Awaish Kumar: Basically approved for the… for the candidates.

11 00:00:50.740 00:00:53.259 Awaish Kumar: So, I suggested only one.

12 00:00:53.380 00:00:55.300 Awaish Kumar: What do you think about that?

13 00:00:56.230 00:01:05.480 Samuel Roberts: Yeah, I was looking at that earlier. I… I wanted to pull it up real quick and have it in front of me, but I… I kind of… I mean… so I… I went a few different ways with it. I like the… there we go, thank you.

14 00:01:05.480 00:01:07.180 Awaish Kumar: It’s on the like, challenge.

15 00:01:07.180 00:01:11.729 Samuel Roberts: Yeah, no, I didn’t have the zoom up, so I was trying to get to the right thing. Yeah.

16 00:01:15.850 00:01:17.330 Samuel Roberts: Yeah, improvement.

17 00:01:17.370 00:01:18.220 Rico Rejoso: Hi.

18 00:01:19.880 00:01:21.139 Samuel Roberts: How are you, Rico?

19 00:01:21.420 00:01:23.349 Rico Rejoso: I’m good, I’m good. Thank you so much for attending.

20 00:01:23.350 00:01:27.470 Awaish Kumar: Like… I think, Rico, you are sharing the screen, right?

21 00:01:27.720 00:01:28.609 Rico Rejoso: Yep, yep.

22 00:01:29.610 00:01:33.049 Awaish Kumar: Yeah, can you look at, like, when I scroll down to Challenge 4?

23 00:01:33.680 00:01:35.550 Rico Rejoso: Yep, the one that you proposed.

24 00:01:36.070 00:01:41.739 Awaish Kumar: Yeah, that’s the only one I propose, so, like, I want, Sam’s feedback on this one.

25 00:01:44.280 00:01:47.250 Samuel Roberts: Yeah, so the idea is…

26 00:01:48.050 00:01:53.429 Samuel Roberts: We need to have them… they need to search this data for something like this chemical formula, basically.

27 00:01:53.800 00:01:55.010 Samuel Roberts: Is that the idea here?

28 00:01:55.010 00:01:56.089 Awaish Kumar: Trying to figure out…

29 00:01:56.320 00:01:57.350 Samuel Roberts: Unstructured data.

30 00:01:57.350 00:02:03.860 Awaish Kumar: Personally, we can… we can, like, I don’t know, I… this was… I remember it from one of the clients.

31 00:02:04.130 00:02:05.080 Samuel Roberts: Okay.

32 00:02:05.410 00:02:14.009 Awaish Kumar: came maybe before you came and Bradford, but we never, like, moved ahead with that client.

33 00:02:14.330 00:02:15.200 Samuel Roberts: Oh, okay.

34 00:02:15.200 00:02:19.159 Awaish Kumar: came as a request, and then it was this. I like this idea.

35 00:02:19.430 00:02:20.530 Samuel Roberts: Yeah, I like it too.

36 00:02:20.530 00:02:24.550 Awaish Kumar: It’s stuck in my mind. So that’s why I proposed it.

37 00:02:24.600 00:02:27.209 Samuel Roberts: It’s not bad, yeah. Do we have all that…

38 00:02:27.210 00:02:40.609 Awaish Kumar: If you think… if you think, like, the data, if, for example, I… we… we, simulate some data, like, we get some PDFs for products, maybe some,

39 00:02:43.090 00:02:56.679 Awaish Kumar: some images, some PDFs, some structured data, and load it in a, like, data folder on GitHub, and then we have a PDF, which basically says the task, deliverables, and things like that.

40 00:02:57.020 00:03:00.300 Awaish Kumar: And we push it to the GitHub as well.

41 00:03:00.300 00:03:05.040 Samuel Roberts: Yeah. So what do you think? Like, if I’m providing… like, if we are providing all the data, like.

42 00:03:05.040 00:03:11.160 Awaish Kumar: The kind of, the, like, the, like, for example, list of 10, 20 products.

43 00:03:11.270 00:03:19.689 Awaish Kumar: which are coming from different, kind of files. Can be PDF, can be CSV, can be,

44 00:03:19.820 00:03:24.789 Awaish Kumar: an image or whatever, and then I provide a list of,

45 00:03:25.030 00:03:32.029 Awaish Kumar: ingredients, which are, like, basically forbidden, like, formulas or whatever.

46 00:03:32.220 00:03:42.979 Awaish Kumar: and the chemical name, like, the… of some things, and maybe I can have a, like, list, like, in a CSV file, and if we give that along with the

47 00:03:43.420 00:03:47.399 Awaish Kumar: Along with the… Like, the task itself.

48 00:03:47.630 00:03:56.629 Awaish Kumar: Do you think it’s enough challenge for them to read it from these files, and basically use AI, to…

49 00:03:57.430 00:04:01.349 Awaish Kumar: Do whatever, like, to give… basically reject or accept the product.

50 00:04:01.850 00:04:04.759 Awaish Kumar: Yeah, I think, like, enough to test their capabilities.

51 00:04:04.760 00:04:11.179 Samuel Roberts: It could be, especially if we have, like, if we… if they know that we’re gonna give them, like, tested on files that we’re not giving them to.

52 00:04:12.460 00:04:13.549 Samuel Roberts: You know what I mean?

53 00:04:14.600 00:04:32.649 Samuel Roberts: So, like, if they know that they need to make it general enough, like a working RESTful API, like you said, then I think that does make sense. It’s very open-ended, which can be good and bad. I like it, because I think it basically is, like, use whatever tools you think are good, just provide us with an endpoint, you know, and, like, the code.

54 00:04:32.650 00:04:33.050 Awaish Kumar: Yeah.

55 00:04:33.050 00:04:48.079 Samuel Roberts: I think it’s a good… yeah, the… what I was kind of struggling with was trying to figure out, like, how complex versus, you know, open-ended to make these things. I think this is a pretty good balance here. My only question is, like, getting all that data together, having it all, like.

56 00:04:48.240 00:04:53.599 Samuel Roberts: you know, making sure… that’s the only, like, I don’t know what we have for that. But yeah, I think that’s not a bad one.

57 00:04:53.600 00:05:00.059 Awaish Kumar: We have to basically search We maybe have to, like, visit some pharmacies, get some, like.

58 00:05:00.190 00:05:04.470 Awaish Kumar: Medic… medical products or something, and a list of

59 00:05:04.700 00:05:18.130 Awaish Kumar: and we just make a list of some forbidden formulas, which are listed on those products somewhere, and that’s all, right? So I can, like, get maybe 50 products, and all the

60 00:05:18.600 00:05:29.769 Awaish Kumar: And the list of all these ingredients on these, which are forbidden, and we can give, like, 20 as a sample to the client… to the candidate, and…

61 00:05:29.820 00:05:39.060 Awaish Kumar: keep the 30 for us to test it afterwards, and yeah, maybe have a CSV, which says accept and rejected, things like that, so we can fully test

62 00:05:39.620 00:05:42.209 Awaish Kumar: Testable? Yeah. Yeah.

63 00:05:42.480 00:05:43.109 Samuel Roberts: I may want to add…

64 00:05:43.110 00:05:45.850 Awaish Kumar: But I would need help figuring out all these things, because…

65 00:05:45.850 00:05:54.490 Samuel Roberts: Yeah, exactly, exactly. Yeah, I may want to add a little bit of, like, because I think we talked about them doing a Loom video for this sort of stuff, to see how well they can talk through their solutions and things.

66 00:05:55.970 00:06:15.419 Awaish Kumar: But I think… Yeah, but I need your help, like, I know they have to write a REST API, they have to give us an endpoint, maybe that requires some kind of coding, but on the AI side, is it enough AI usage, or is it enough complexity to test the candidate? I, like.

67 00:06:15.490 00:06:19.010 Awaish Kumar: Is it enough? Also, it is not too big, right? Like…

68 00:06:19.190 00:06:27.980 Awaish Kumar: it’s like, something can be… can be done in, like, maybe in 5 to 10 hours, like, we don’t want them to spend, like,

69 00:06:28.170 00:06:29.610 Awaish Kumar: Days on it, like, it should.

70 00:06:29.610 00:06:31.060 Samuel Roberts: No, no, I agree, I agree.

71 00:06:31.060 00:06:32.070 Awaish Kumar: 5 hours.

72 00:06:32.440 00:06:39.829 Samuel Roberts: I think… I think 5 hours seems reasonable for this. You know, obviously that depends on the applicant, and so part of it is trying to match, you know, what they’re.

73 00:06:39.830 00:06:42.449 Awaish Kumar: No, no, like, I think it’s fine.

74 00:06:42.650 00:06:43.209 Awaish Kumar: For humans?

75 00:06:43.540 00:06:47.000 Awaish Kumar: template. Like, then it should be under 5 hours.

76 00:06:47.180 00:06:48.299 Samuel Roberts: No, you’re absolutely right.

77 00:06:50.940 00:06:54.729 Samuel Roberts: Yeah, I think it’s probably a smart thing to do. I think that’s not bad, yeah.

78 00:06:55.300 00:07:02.440 Samuel Roberts: But yeah, getting all that stuff together, maybe running a dry run, or when we give it out to someone, I would do it as well, or something, yeah.

79 00:07:03.380 00:07:10.980 Awaish Kumar: That was just I wanted feedback just because I propose it, but now if you have… if you think

80 00:07:11.210 00:07:16.560 Awaish Kumar: Any of the… I just do have proposals, it’s good enough, like, we can discuss that as well.

81 00:07:16.840 00:07:19.950 Samuel Roberts: I mean, the other ones were a little more, like,

82 00:07:20.510 00:07:25.420 Samuel Roberts: structured, and I actually like… I was trying to make them a little more open.

83 00:07:25.910 00:07:35.009 Samuel Roberts: So I like… I like the openness of it. Maybe we just need to, like, flesh it out a little bit with, obviously, the… the files and stuff, and maybe give a little bit more of, like, a…

84 00:07:35.420 00:07:37.620 Samuel Roberts: You know.

85 00:07:37.730 00:07:54.009 Samuel Roberts: options for, like, how they present it to us. Like, is it just an API that we can use? Is it gonna… are they gonna have a UI at all? Are they gonna do anything like that? Like, you know, we don’t necessarily want to point them to, like, which tools to use, but I want them to make those decisions themselves a little bit.

86 00:07:55.650 00:08:01.229 Samuel Roberts: But yeah, I think… I think that’s not a bad one, and that feels a little more real-world than the ones I was kind of contriving, so…

87 00:08:01.230 00:08:02.140 Awaish Kumar: Yeah, yeah.

88 00:08:02.430 00:08:04.559 Awaish Kumar: That’s the real client work.

89 00:08:04.560 00:08:08.969 Samuel Roberts: Yeah, exactly, that seems great. I think that’s a good one. And then, because, like, it’s not…

90 00:08:09.270 00:08:27.439 Samuel Roberts: so complex that it’s gonna be very, you know, long, but it’s not so simple that I think, you know, if someone’s very experienced, they might knock it out of the park easily, and we’ll be able to see that. And if someone’s not, they might struggle more, and we’ll see. You know, we want to see that stuff, we don’t just want them to, like, crush it, you know?

91 00:08:27.810 00:08:29.369 Samuel Roberts: So I think that’s not a bad one.

92 00:08:29.540 00:08:31.459 Samuel Roberts: I think we, yeah, we probably just need to…

93 00:08:31.660 00:08:34.710 Samuel Roberts: Source some of that stuff, probably.

94 00:08:34.710 00:08:35.230 Awaish Kumar: I, I’m…

95 00:08:35.230 00:08:36.600 Samuel Roberts: Yeah.

96 00:08:36.600 00:08:43.160 Awaish Kumar: I’m just… I want to understand, like, I… I can basically… I’ll push them in the…

97 00:08:43.580 00:08:52.089 Awaish Kumar: and, like, all those PDFs, images in ChatGPT get the answers, so I just want, like, it’s not very simple, like.

98 00:08:52.250 00:08:56.360 Awaish Kumar: In, like, you can do it in 10 minutes, you know, or maybe 30 minutes.

99 00:08:56.710 00:09:02.519 Awaish Kumar: Well, that’s what I’m saying, I think, like, we want to make sure that they know that they need to build something that’s more usable beyond just, you know.

100 00:09:02.560 00:09:08.950 Samuel Roberts: like, yeah, you could put a PDF into something, but, like, they need to build something that’s gonna properly ingest it, do it, because then we’re gonna.

101 00:09:08.950 00:09:09.569 Awaish Kumar: defeated one.

102 00:09:09.570 00:09:10.210 Samuel Roberts: that aren’t…

103 00:09:10.210 00:09:17.600 Awaish Kumar: share properly, properly extract it, and then send this information to some API, along with a list of forbidden

104 00:09:17.780 00:09:24.080 Awaish Kumar: Exactly, yeah. Come to formulas as context, and then get the answer if it is rejected or accepted.

105 00:09:24.080 00:09:31.980 Samuel Roberts: Right, so yeah, so, like, it’s a big ingestion kind of question there, and then depending on if they just want to do pure LLM, or they want to build more agentic stuff.

106 00:09:33.720 00:09:49.539 Samuel Roberts: you know, they might go a few different ways with it, but it’s not… yeah, see, it’s not crazy. The ingestion is one thing, the actual prompting is going to be another thing, how we’re able to access it, whether it’s an API or, you know, they build some UI or something, is the other side of it, which I think is good.

107 00:09:49.540 00:09:53.849 Awaish Kumar: We can’t even keep that… we can’t even keep that as open-ended, like…

108 00:09:53.850 00:09:55.049 Samuel Roberts: Exactly, I want to see what they do.

109 00:09:55.490 00:09:56.399 Samuel Roberts: We’ll see what they do, yeah.

110 00:09:56.400 00:09:57.620 Awaish Kumar: I or UI, whatever.

111 00:09:57.620 00:09:59.729 Samuel Roberts: Exactly. I think that’s good to keep open.

112 00:10:00.290 00:10:02.460 Samuel Roberts: Cause then we’ll see, like, how…

113 00:10:02.580 00:10:15.639 Samuel Roberts: how in-depth they go, how quick they go. You know, if they do something quick and dirty and it works, that’s one thing. If they do something very long and complicated and it works, that’s different. If it doesn’t work, it’s different. You know, there’s a few different ways they can go with it, so I like that.

114 00:10:15.640 00:10:16.280 Awaish Kumar: Okay.

115 00:10:18.070 00:10:29.980 Awaish Kumar: Okay, then the next… if we are both okay with that, then, like, we might have to go further to the next steps, is the identification, like, getting the, actually, the raw data?

116 00:10:30.270 00:10:37.700 Awaish Kumar: And, Collecting the… A few different products, and a list of ingredients.

117 00:10:37.830 00:10:44.520 Awaish Kumar: Yeah. Like, they have to… they have to get the ingredients, basically, but we just provide the… the forbidden ones, like…

118 00:10:44.620 00:10:50.330 Awaish Kumar: Right. A product might have 100, like, 10, 20 ingredients on its image.

119 00:10:50.470 00:10:56.410 Awaish Kumar: But we can say, okay, only the one which has, Glucose is forbidden, for example.

120 00:10:56.410 00:10:57.620 Samuel Roberts: Right, right.

121 00:10:57.750 00:11:02.160 Samuel Roberts: We might even want to make… tell them that, like, we want to be able to do this with a different… multiple different lists.

122 00:11:02.520 00:11:03.400 Samuel Roberts: You know?

123 00:11:05.200 00:11:14.260 Samuel Roberts: make that a variable that they have to account for, that we input what we don’t want it to be, so they can’t just, you know, hardcode things and just run it through ChatGPT kind of thing.

124 00:11:16.810 00:11:24.750 Awaish Kumar: Yeah, yeah, like, we can sell them what we don’t need, but, like, they… they obviously have to take care of the

125 00:11:24.990 00:11:29.730 Awaish Kumar: You know, different formatting, or different… Like, if I’m,

126 00:11:29.890 00:11:33.759 Awaish Kumar: The name is, like, the formula is, like, written.

127 00:11:33.760 00:11:34.080 Samuel Roberts: Yep.

128 00:11:34.080 00:11:36.620 Awaish Kumar: A little bit differently, or something like that?

129 00:11:36.620 00:11:43.530 Samuel Roberts: Yeah, yeah, yeah. Yeah, the name versus the chemical formula versus… yeah, that’s true, that’s true. I like that.

130 00:11:43.690 00:11:44.560 Samuel Roberts: Okay.

131 00:11:45.480 00:11:47.880 Samuel Roberts: So I think…

132 00:11:48.290 00:11:53.670 Samuel Roberts: We just need to get basically a list of products together, or, data on products.

133 00:11:55.370 00:12:00.139 Awaish Kumar: Yeah, step one is, getting the list of products, or, like…

134 00:12:00.260 00:12:08.399 Awaish Kumar: List of product means all these, in a… in some kind of file, like, it can be an image, it can be a PDF, it can be a…

135 00:12:08.630 00:12:11.220 Awaish Kumar: just a text file with the information. Of course, of course

136 00:12:11.580 00:12:21.619 Awaish Kumar: Yeah, and then, if we have maybe more than 20, 30, that could be enough. Then we just need a PDF, which describes the…

137 00:12:22.740 00:12:26.500 Awaish Kumar: Just describe the task and deliverables in detail.

138 00:12:26.500 00:12:29.219 Samuel Roberts: I think we can just make that a README file in the repo, probably.

139 00:12:29.220 00:12:30.590 Awaish Kumar: Yeah, okay, yeah.

140 00:12:30.590 00:12:33.649 Samuel Roberts: And then they… and then basically they’ll make, like, a PR…

141 00:12:33.770 00:12:36.470 Samuel Roberts: To that repo with their code.

142 00:12:36.640 00:12:39.629 Samuel Roberts: And then a Loom video, maybe explaining it.

143 00:12:39.940 00:12:42.030 Samuel Roberts: And that’s their submission, basically.

144 00:12:44.160 00:12:44.870 Awaish Kumar: Okay.

145 00:12:47.990 00:12:53.259 Samuel Roberts: Okay, cool. So I guess I don’t know best where to gather that data from.

146 00:12:54.640 00:12:58.200 Awaish Kumar: I… I’m just going to go to the internet and start.

147 00:12:58.200 00:13:04.860 Samuel Roberts: Okay, that’s what, yeah, okay. If you want to start doing that, I can put together a README and make a repo that we can start.

148 00:13:06.350 00:13:08.640 Awaish Kumar: Okay, okay, sure.

149 00:13:09.160 00:13:15.540 Samuel Roberts: And then if you need help finding things, let me know, like, see what you can find and how many are out there, because, yeah, we won’t…

150 00:13:15.540 00:13:16.510 Awaish Kumar: everything in the repo.

151 00:13:16.510 00:13:17.190 Samuel Roberts: But, yeah.

152 00:13:17.190 00:13:20.089 Awaish Kumar: But it doesn’t need to be today, right?

153 00:13:20.760 00:13:23.040 Samuel Roberts: Yeah, Rico, what is the timeline on this?

154 00:13:23.040 00:13:25.619 Awaish Kumar: I don’t have, like, enough time today.

155 00:13:25.620 00:13:29.180 Samuel Roberts: Yeah, I don’t think today’s gonna be it. No, no, no, I don’t think sake.

156 00:13:29.500 00:13:36.099 Rico Rejoso: I mean, maybe, I don’t know, tomorrow, how’s your schedule for tomorrow?

157 00:13:38.030 00:13:41.279 Awaish Kumar: Oh, straight, okay, I can,

158 00:13:41.500 00:13:51.880 Awaish Kumar: start on it tomorrow, gather a few products, and then we can see, like, maybe, by early next week, Monday, Tuesday, we’ll have complete…

159 00:13:51.980 00:13:58.070 Samuel Roberts: We could have it pretty complete, a repo with, you know, half the files in it, we’ll keep half for us, and then…

160 00:13:58.070 00:13:58.860 Awaish Kumar: Huh, nope.

161 00:13:58.860 00:14:03.110 Samuel Roberts: Yeah, then we can share that, and the README will have everything that they need.

162 00:14:03.840 00:14:04.400 Rico Rejoso: mate.

163 00:14:04.820 00:14:05.470 Samuel Roberts: Cool.

164 00:14:06.220 00:14:07.469 Rico Rejoso: Let me…

165 00:14:07.470 00:14:08.170 Awaish Kumar: The one that…

166 00:14:08.170 00:14:09.640 Rico Rejoso: The chat.

167 00:14:09.640 00:14:11.949 Awaish Kumar: There’s a third point also.

168 00:14:12.840 00:14:13.520 Rico Rejoso: Go ahead, yes.

169 00:14:13.520 00:14:18.289 Awaish Kumar: Gather raw data, and then also the… the number two for me is also to

170 00:14:18.550 00:14:24.150 Awaish Kumar: Make a list of forbidden, Part of, like, the ingredients.

171 00:14:25.700 00:14:26.090 Samuel Roberts: Yeah.

172 00:14:26.090 00:14:26.790 Rico Rejoso: make…

173 00:14:28.460 00:14:31.579 Awaish Kumar: Orbital ingredients. Ingredients.

174 00:14:32.950 00:14:33.710 Rico Rejoso: Okay.

175 00:14:34.880 00:14:37.740 Awaish Kumar: I don’t know what to name, like, what to say, like, chemical.

176 00:14:37.740 00:14:51.800 Samuel Roberts: Yeah, I think… I think that’s… or just, like, yeah, like, something… we can even make it more general, where it’s just, like, if it does, you know… I think forbidden ingredients is a really good way to put it for now, but, like, even if there’s something else where it’s like, this one has this side effect, reject it, you know? Or anything like that.

177 00:14:51.800 00:14:52.569 Awaish Kumar: Oh, yeah.

178 00:14:52.570 00:14:55.009 Samuel Roberts: We could broaden it if we need to. So, I would say the first thing.

179 00:14:55.010 00:15:00.280 Awaish Kumar: I think the thing is, like, we try to just keep it simple. We don’t, like…

180 00:15:00.280 00:15:00.780 Samuel Roberts: That’s true.

181 00:15:00.780 00:15:04.050 Awaish Kumar: No, no, no, like… Releasing it for a client right now, like.

182 00:15:04.050 00:15:04.690 Samuel Roberts: Right, right.

183 00:15:04.690 00:15:07.099 Awaish Kumar: We can do it, and that’s all.

184 00:15:07.240 00:15:08.010 Samuel Roberts: Yeah, okay, that’s good.

185 00:15:08.710 00:15:13.110 Rico Rejoso: So, we’re going with the challenge for product safety compliance AI, right?

186 00:15:14.350 00:15:19.860 Samuel Roberts: Yes, I think so. I like that it’s a real client work, because I was just kind of making stuff up that I, you know, I like…

187 00:15:19.860 00:15:20.460 Rico Rejoso: Affinity.

188 00:15:20.460 00:15:21.010 Awaish Kumar: Yep.

189 00:15:21.500 00:15:29.599 Rico Rejoso: Yeah, I was thinking of just going with one challenge. Just to give you an idea, here’s the form that they will be filling out, and…

190 00:15:29.600 00:15:30.840 Samuel Roberts: Oh, okay.

191 00:15:31.260 00:15:40.890 Rico Rejoso: previously we’re just doing, or we just, you know, wanted them to provide a Loom video, answering these questions, but since there’s been, like, you know.

192 00:15:40.970 00:15:58.889 Rico Rejoso: effective results from the previous applicants or onboardies that we have, come up with the idea of giving them technical, stuff to work on to gauge their, capabilities and experience, based on the challenges that you guys will be providing.

193 00:15:58.890 00:16:00.930 Samuel Roberts: Oh, okay. So is this going…

194 00:16:01.330 00:16:04.359 Samuel Roberts: Sorry, so just flip back over to that application real quick.

195 00:16:04.940 00:16:10.029 Samuel Roberts: So this is, like, the first round? Like, this is what they were gonna fill out on the website or something?

196 00:16:11.400 00:16:26.599 Rico Rejoso: Yeah, once they click on that job or role, they will be given this form to fill out, and one thing would be the Loom video, which I think wasn’t very effective for us to gauge their, you know, experience and capabilities.

197 00:16:26.600 00:16:27.180 Samuel Roberts: Sure.

198 00:16:27.180 00:16:28.430 Rico Rejoso: Doing stuff for brainstorm.

199 00:16:28.430 00:16:32.999 Samuel Roberts: So this is… this is not a second screening, this is a part of the first screening.

200 00:16:33.000 00:16:37.339 Rico Rejoso: Yeah, so I’ll be changing this one and putting in the challenge here.

201 00:16:37.480 00:16:43.100 Rico Rejoso: this one, so I also need to gather the deliverables, so I can create, you know.

202 00:16:43.100 00:16:48.110 Samuel Roberts: Yeah, okay, hold on, I didn’t realize this was… I thought this was gonna be something we’d be sharing on a repo with people.

203 00:16:48.710 00:16:50.070 Samuel Roberts: after the fact.

204 00:16:50.230 00:16:55.830 Samuel Roberts: Can you share this… actually, can I see this, applicant’s file? Can you share that link, too?

205 00:16:58.600 00:17:00.400 Rico Rejoso: Because my thought was that…

206 00:17:00.620 00:17:04.580 Samuel Roberts: There’d be, like, a screening stage like this, and then a tech screening.

207 00:17:07.220 00:17:13.039 Rico Rejoso: Yeah, like… Okay, let me… What we are talking about is a complete…

208 00:17:13.109 00:17:15.009 Awaish Kumar: Complete, like, a project.

209 00:17:15.010 00:17:22.550 Samuel Roberts: Yeah, this is, like, a bigger, you know, because, like, this is an application, but, like, this is, like, a technical thing we give people that pass this first round, is what I thought.

210 00:17:24.000 00:17:35.410 Rico Rejoso: Okay, so the challenge that you come up with is more like of a second round kind of thing, where if they made it through this first, or this initial application stage, they proceed on

211 00:17:35.660 00:17:41.930 Rico Rejoso: Fundalizing the, technical stuff, right?

212 00:17:42.740 00:17:43.270 Samuel Roberts: Yeah, I think…

213 00:17:43.270 00:17:46.580 Rico Rejoso: challenge that you created. I’m sorry, I wasn’t able to get everything.

214 00:17:46.580 00:17:57.099 Samuel Roberts: Yeah, I mean, I think we’d probably ask them to time box it at no more than, you know… I mean, I think… wait, she said, like, 5 today. I think 5 is probably, like, more… you know, we don’t want them… we want to time box it to, like…

215 00:17:57.100 00:18:03.610 Awaish Kumar: We need to give them the timeline. We can say that this challenge is going to take you around

216 00:18:03.740 00:18:10.679 Awaish Kumar: 5 hours max, but we need to give them time. We can’t give them 5 hours to do that. We have to give, like… No, no, no.

217 00:18:10.680 00:18:16.260 Samuel Roberts: That’s what I’m saying, I think, I think, yeah, we would make this, like, a… say it’s, like, a 5-hour challenge,

218 00:18:17.430 00:18:20.129 Samuel Roberts: At most… at max 5-hour channels, put it that way.

219 00:18:20.460 00:18:21.650 Awaish Kumar: Yep, yep.

220 00:18:24.200 00:18:29.599 Samuel Roberts: Which, again, I think might be a little, you know, depending on the applicant, might be overkill, but…

221 00:18:29.710 00:18:34.170 Samuel Roberts: I don’t… I don’t see this being 5 hours, but I like that.

222 00:18:34.840 00:18:35.500 Samuel Roberts: Exactly.

223 00:18:35.500 00:18:42.459 Awaish Kumar: depending on the application, like, applicant, it doesn’t even take, like, maybe for AI experts, like, it would take.

224 00:18:42.460 00:18:43.860 Samuel Roberts: Exactly, that’s what I’m saying. I think…

225 00:18:44.670 00:18:49.299 Samuel Roberts: at 5 is fine, but this is definitely a second round kind of thing, not a first application thing.

226 00:18:49.300 00:18:49.670 Rico Rejoso: Okay.

227 00:18:49.670 00:18:50.010 Awaish Kumar: Oh, no.

228 00:18:50.010 00:18:52.829 Rico Rejoso: I can have that, that’s fine, that’s fine, we can post that.

229 00:18:52.990 00:19:02.789 Rico Rejoso: So we’ll just stick with the first kind of form, the initial one, which is submitting a Loom video. This is a different one where they have… I mean, we’ll be testing their technical skills.

230 00:19:03.100 00:19:05.309 Rico Rejoso: Yeah, that’s fine. We can have this.

231 00:19:05.310 00:19:16.540 Awaish Kumar: Yeah, I don’t know if we want to, like, first screen could be… if we really want someone to do some coding on our first screen, it can be something like Hecker Act, right?

232 00:19:16.690 00:19:21.850 Awaish Kumar: Where you can send them a link of some… yeah, some coding challenge in Python.

233 00:19:21.850 00:19:24.859 Samuel Roberts: Yeah, that’s more… that’s more for an initial screen, I think, definitely.

234 00:19:24.860 00:19:25.730 Awaish Kumar: Oh, yeah.

235 00:19:26.110 00:19:28.060 Awaish Kumar: That is the initial screen.

236 00:19:28.420 00:19:45.510 Rico Rejoso: Well, I guess the… I mean, my idea was, I don’t want… because it’s with Thomas doing the interview and everything, and Amber, who’s, screening those applicants, and I don’t want them to spend so much time on one applicant from the first round to the second round, knowing the second round that they will fail.

237 00:19:45.940 00:19:55.599 Samuel Roberts: Okay, that’s… that’s smart. Okay, yeah, so what is… actually, that’s… that’s helpful. Can we back up a step here? Like, what is the normal application process right now? They fill out this application, including all the information…

238 00:19:55.600 00:20:02.090 Rico Rejoso: Yeah, Robert or Utham would be gauging them or watching the video, the Loom video, and

239 00:20:02.230 00:20:06.320 Rico Rejoso: Based on that, we’ll come up a decision if they would proceed to the second round, which is…

240 00:20:06.570 00:20:19.520 Rico Rejoso: For data, Robert will be doing the interview for any AI-related interview, or applicants who come would be doing the interview, and they’ll be asking their own specific questions for the specific… for the role or position.

241 00:20:19.930 00:20:24.770 Rico Rejoso: And after that, there will be a final interview, which is technically the job offer.

242 00:20:24.970 00:20:27.969 Rico Rejoso: So there’s, like, a three-stage process that we’re doing right now.

243 00:20:28.230 00:20:28.850 Samuel Roberts: Oh, okay.

244 00:20:28.850 00:20:29.740 Rico Rejoso: I think about that one.

245 00:20:29.740 00:20:30.920 Samuel Roberts: Go ahead, yeah, sorry, continue.

246 00:20:30.920 00:20:39.939 Rico Rejoso: depending on that one, actually, we’re trying to incorporate, you guys, Sam and Awish, to do the interview for data and AI applicants.

247 00:20:39.940 00:20:40.800 Awaish Kumar: Yeah, we need…

248 00:20:40.800 00:20:41.749 Rico Rejoso: loop as well.

249 00:20:42.790 00:20:44.710 Samuel Roberts: Right. But if in the first round.

250 00:20:44.730 00:20:57.249 Rico Rejoso: Utam is satisfied with the experience and the way that applicants perform. We usually, go on with the job offer and, you know, then won’t…

251 00:20:57.680 00:21:01.660 Rico Rejoso: Have them go through interviewing you guys, or doing an interview with you guys.

252 00:21:01.660 00:21:05.200 Samuel Roberts: Okay. To save time, and depending on the urgency of the position.

253 00:21:06.100 00:21:09.219 Awaish Kumar: Yeah, I think the… Now the flow should be, like…

254 00:21:09.830 00:21:15.259 Awaish Kumar: We get the first responses and the Loom video. Utama Robert reviews it.

255 00:21:15.780 00:21:18.820 Awaish Kumar: And then, filters out the candidates.

256 00:21:19.090 00:21:20.460 Awaish Kumar: For the challenge?

257 00:21:20.600 00:21:23.559 Samuel Roberts: Then we send them the challenge, and .

258 00:21:23.880 00:21:27.950 Awaish Kumar: After they complete the challenge, we book a, like, a technical call with them.

259 00:21:28.130 00:21:28.660 Samuel Roberts: Yep.

260 00:21:28.660 00:21:33.400 Awaish Kumar: So we reviewed this channel together with him, and along with technical questions.

261 00:21:33.600 00:21:35.529 Awaish Kumar: And that would be, like, kind of…

262 00:21:35.850 00:21:42.240 Awaish Kumar: Second round, and then third is the, I mean, maybe the offer, or meeting with Utam, or something.

263 00:21:42.240 00:21:45.050 Samuel Roberts: Yeah, I think that makes more sense to me. That way…

264 00:21:45.050 00:21:48.919 Rico Rejoso: What was the… what was the suggested process? Were they able to get… get all that?

265 00:21:48.920 00:21:52.219 Awaish Kumar: Yeah, the first Loom video, like, the application.

266 00:21:52.500 00:21:53.210 Rico Rejoso: Huh.

267 00:21:53.870 00:21:58.350 Awaish Kumar: And second is sending the challenge, if they pass the first round.

268 00:21:58.740 00:22:01.910 Awaish Kumar: The second round is sending them the challenge.

269 00:22:02.790 00:22:03.410 Rico Rejoso: Okay.

270 00:22:03.890 00:22:09.340 Awaish Kumar: But that is, like, second round. Like, first round is they fill the form, they send a, like, rule video.

271 00:22:09.960 00:22:17.809 Awaish Kumar: Second round is, we send them a challenge, and they submit the response. Third is a technical screen.

272 00:22:18.130 00:22:33.449 Awaish Kumar: Where we discuss this challenge, how they implemented the solution, they… like, why they think, like, that was the best… best way to implement it. And we can also go further… deep into their expertise. Like, the third round is this.

273 00:22:33.730 00:22:41.079 Awaish Kumar: And the fourth one could be, interview with, like, the final 30-minute

274 00:22:41.350 00:22:45.339 Awaish Kumar: interview with Utam or Robert, or maybe directly the offer.

275 00:22:47.900 00:22:48.650 Rico Rejoso: Okay.

276 00:22:48.760 00:22:49.909 Rico Rejoso: I like this one.

277 00:22:50.080 00:22:53.800 Rico Rejoso: So they have to go through with you guys first before they went to UTOM.

278 00:22:54.920 00:23:01.599 Samuel Roberts: Yeah, like, I think there’s still this initial loom screening, might be UTAM or Robert still, whoever that is.

279 00:23:01.810 00:23:02.799 Samuel Roberts: Yeah, exactly, but it’s.

280 00:23:02.800 00:23:08.240 Awaish Kumar: That depends on who makes the decision. Exactly, but I think… That can be Rico, or Utam, or anyone.

281 00:23:08.390 00:23:15.209 Samuel Roberts: Exactly, but I think, yeah, so they at least get a look at everyone, then they send them the tech challenge, and then we review the tech challenge.

282 00:23:15.570 00:23:24.220 Awaish Kumar: Yeah, same goes for tech challenge as well, like, if we review it, if it is, like, it doesn’t seem to be working, we don’t, like, invite them for a tech screen.

283 00:23:24.220 00:23:27.639 Samuel Roberts: Exactly, yeah, every one of those is a… is a, like, go-no-go decision.

284 00:23:27.640 00:23:28.260 Awaish Kumar: Yeah.

285 00:23:28.810 00:23:29.719 Rico Rejoso: Got it, got it.

286 00:23:30.650 00:23:35.130 Rico Rejoso: Okay, this… so this could be either Awish or Sam.

287 00:23:35.280 00:23:35.820 Samuel Roberts: Yep.

288 00:23:36.110 00:23:53.680 Rico Rejoso: this one also. So we’re gonna… we’re gonna create a set of challenge, or, I mean, a challenge for… this is for AI… for AI applicants, right? So we’re gonna also create one for the data applicants, which should be coming from you guys as well, right? So I think you’ll all be prepared with that.

289 00:23:54.270 00:23:58.259 Samuel Roberts: Yeah, yeah. Yeah, and then we’ll probably split these that way, too, I imagine, for the applicants, so…

290 00:23:59.460 00:23:59.900 Rico Rejoso: Okay.

291 00:23:59.900 00:24:00.550 Samuel Roberts: Cool.

292 00:24:02.030 00:24:09.480 Rico Rejoso: let me get this one, and also, I wasn’t able to… I mean, the next step would be gather the raw data.

293 00:24:09.590 00:24:13.549 Samuel Roberts: Yep. Make a list of forbidden ingredients. Sam would be creating a repo.

294 00:24:13.980 00:24:14.350 Samuel Roberts: Yep.

295 00:24:14.350 00:24:27.960 Rico Rejoso: once we get that one, we can also… we might also create a form, or a Notion form, that would contain all of this. What do you think is the best way to provide all the details, or for the challenge?

296 00:24:28.610 00:24:33.540 Samuel Roberts: Well, I think once we create the repo, we’ll be able to share that two people.

297 00:24:34.060 00:24:35.380 Rico Rejoso: And so then… Oh, that way.

298 00:24:35.380 00:24:38.190 Samuel Roberts: That… all the README will have all the information.

299 00:24:38.280 00:24:56.109 Samuel Roberts: kind of like what we just discussed and all this stuff here for Challenge 4, with a little bit more, you know, maybe a little more context, pointing to the files, explaining, you know, Python or TypeScript or something, however we want to do that. But we can go back and forth with that a little bit. And then that’ll explain to do the work, make a PR,

300 00:24:56.210 00:24:58.089 Samuel Roberts: And then that’ll be their submission.

301 00:24:59.100 00:25:00.300 Rico Rejoso: Okay, got it.

302 00:25:01.070 00:25:03.730 Rico Rejoso: we can proceed with this one. I think…

303 00:25:05.430 00:25:10.120 Rico Rejoso: Sounds good. The technical challenge, which would be a repo.

304 00:25:10.330 00:25:10.870 Rico Rejoso: to be safe.

305 00:25:10.870 00:25:11.210 Samuel Roberts: Yes.

306 00:25:11.210 00:25:15.159 Rico Rejoso: Which we can send via, through link, via email, right?

307 00:25:15.520 00:25:19.760 Samuel Roberts: Yeah, yeah, so we’ll have it in there. I gotta figure out, cause…

308 00:25:20.010 00:25:24.850 Samuel Roberts: Sometimes I’ve done it where it’s just a public repo, but I think we want… We don’t want people.

309 00:25:24.850 00:25:26.489 Awaish Kumar: We’ll see who do people’s…

310 00:25:26.490 00:25:29.260 Samuel Roberts: Yeah, so we gotta keep it private, so we might need to know their GitHub.

311 00:25:29.380 00:25:32.110 Samuel Roberts: But…

312 00:25:32.450 00:25:41.429 Samuel Roberts: we might be able to set up a way to either have it so that when they click a link, they get access, or… are they submitting anything in that initial thing with their GitHub, name?

313 00:25:42.160 00:25:43.860 Rico Rejoso: No. Should be included?

314 00:25:44.770 00:25:46.089 Rico Rejoso: We can include it if you want.

315 00:25:46.620 00:25:49.509 Samuel Roberts: Yeah, I think we want to ask them for their GitHub, username.

316 00:25:50.190 00:25:52.799 Samuel Roberts: And that’ll make sending that, sharing that with them easy.

317 00:25:56.960 00:25:58.730 Samuel Roberts: And then also, yeah, make sure that they know…

318 00:25:58.730 00:26:06.179 Awaish Kumar: So anyone, that is our filter also. Anyone who is not using GitHub, like, that’s filtered out.

319 00:26:06.180 00:26:07.129 Samuel Roberts: Yeah, true, true, true.

320 00:26:07.130 00:26:08.050 Rico Rejoso: Okay, yeah.

321 00:26:08.050 00:26:26.450 Samuel Roberts: Yeah, I definitely say, like, add GitHub username, and make, like, a, you know, we will share… this will be to share a technical challenge if you make it to the next stage or something, so they know. Okay. Because sometimes I’m like, I, you know, my GitHub has mostly private stuff on it, so it’s not like I share it all the time, because it’s not really where I see a lot of my work, but make sure they know that we need it for that, yeah.

322 00:26:28.690 00:26:39.640 Rico Rejoso: Once they pass, we will be send… we’ll be adding them to the repo and send them a confirmation through email that they can proceed with a technical challenge, right?

323 00:26:40.750 00:26:41.340 Samuel Roberts: Yeah.

324 00:26:42.360 00:26:46.529 Awaish Kumar: Yeah, and that README will have instruction for how to submit the challenge, basically.

325 00:26:46.530 00:26:47.200 Samuel Roberts: Exactly, yeah.

326 00:26:47.200 00:26:50.199 Awaish Kumar: There’s a new branch submitted to the PR model.

327 00:26:51.000 00:26:55.429 Rico Rejoso: Okay, so how would you guys be informed once they’re done with their…

328 00:26:55.800 00:26:59.349 Rico Rejoso: With the challenge. Should they also send an email confirming.

329 00:26:59.350 00:27:01.440 Samuel Roberts: They would… they would probably…

330 00:27:01.990 00:27:12.110 Samuel Roberts: when they make the PR, we would get an update. PR is, like, a pull request on GitHub, so it’s basically, like, they want to merge their code in, and we would just see that as.

331 00:27:13.450 00:27:13.830 Awaish Kumar: Fair enough.

332 00:27:13.830 00:27:16.350 Samuel Roberts: We would see that in GitHub and get an email that way, so…

333 00:27:18.740 00:27:21.740 Rico Rejoso: I… I’m sorry, I just want to be, you know, specific with the price.

334 00:27:21.740 00:27:24.950 Samuel Roberts: Yeah, yeah, we’ll get notified by GitHub when they make a pull request.

335 00:27:26.360 00:27:28.209 Awaish Kumar: I think that’s the best way to do it, but…

336 00:27:28.470 00:27:35.910 Awaish Kumar: Yeah, we can, like, we can obviously set up the notifications, but also in our email, when we are sending the channels, we can ask them.

337 00:27:36.130 00:27:39.420 Awaish Kumar: Once you’re done, like… Send us the…

338 00:27:39.420 00:27:40.170 Samuel Roberts: We could…

339 00:27:40.170 00:27:41.579 Awaish Kumar: your branch, or with some…

340 00:27:41.580 00:27:47.290 Samuel Roberts: Yeah, we could definitely do that, that might be even better. Yeah, link to the branch, or a fork even, that way they keep them separated from this.

341 00:27:49.330 00:27:51.030 Rico Rejoso: Yeah, we can figure that out, cause that…

342 00:27:51.110 00:27:54.860 Samuel Roberts: I’m not sure the best way to keep it so that they can’t see other people’s submissions, you know?

343 00:27:56.900 00:28:03.210 Awaish Kumar: Yeah, I am not sure either, like, is it possible in GitHub that you can only see the branches you have created.

344 00:28:03.210 00:28:09.830 Samuel Roberts: I was just… I was just googling that to see, because either it’ll be a branch, or we should just ask them to fork it and share the repo.

345 00:28:10.130 00:28:11.350 Awaish Kumar: Yeah, okay.

346 00:28:11.640 00:28:17.390 Samuel Roberts: But I’ll figure that out, and I’ll put that all in the repo, and I’ll make sure to explain that to you, Rico, once we have that sorted out.

347 00:28:17.390 00:28:20.980 Awaish Kumar: Yeah, but Rico, that will all be included in the README, so…

348 00:28:21.570 00:28:25.329 Awaish Kumar: But I’ll make sure that you know the process as well, so yeah.

349 00:28:25.330 00:28:31.309 Rico Rejoso: I will be creating an SOP for this one, so we all… we can, you know, just share the document.

350 00:28:31.310 00:28:31.790 Awaish Kumar: Yo, hold on.

351 00:28:31.790 00:28:32.639 Rico Rejoso: Process and everything.

352 00:28:33.450 00:28:41.740 Awaish Kumar: Okay, once, like, we figure out when the repo is ready, like, you will have everything you need to communicate with the candidate.

353 00:28:41.740 00:28:42.170 Samuel Roberts: Yeah.

354 00:28:42.170 00:28:49.109 Rico Rejoso: Alrighty, that’s great, thank you so much for that, guys. And technical screen with tech leads, so you’ll be confirming with us once they pass.

355 00:28:49.110 00:28:49.770 Awaish Kumar: Yeah, technically.

356 00:28:49.770 00:28:50.810 Rico Rejoso: final interview.

357 00:28:51.920 00:29:01.190 Awaish Kumar: TechScreen will submit the… Submit the final notes to… with the operations team and the Utham Robert.

358 00:29:01.440 00:29:02.030 Samuel Roberts: Yeah.

359 00:29:04.880 00:29:06.440 Rico Rejoso: Okay, fair.

360 00:29:06.740 00:29:18.670 Rico Rejoso: And final interview will be Robert, then job offer. Alright, I think this is a much detailed process that we can have. I’m gonna… I mean, gonna, create a draft for this one, and…

361 00:29:18.670 00:29:27.469 Rico Rejoso: sent to you, Tom. So, we’ll just be waiting for the challenge, or finalized challenge for it. You mentioned around next week, Tuesday?

362 00:29:29.890 00:29:31.170 Awaish Kumar: Yeah, yep.

363 00:29:31.170 00:29:31.810 Rico Rejoso: Okay.

364 00:29:31.950 00:29:45.619 Rico Rejoso: Alrighty. So, I’ll be following up by Monday, or at least if we can provide update by tomorrow, if there is any, then I’ll be following up by Monday for this one. Perfect. Okay? And Tuesday, we’ll be providing the final draft to you, Tom.

365 00:29:47.080 00:29:48.020 Rico Rejoso: Alright.

366 00:29:48.330 00:29:52.100 Rico Rejoso: I think that’s it for me. Anything else that you wanted to discuss or go through it?

367 00:29:52.100 00:29:53.080 Awaish Kumar: Nope.

368 00:29:53.340 00:29:54.930 Samuel Roberts: No, this is good. I think we made good progress.

369 00:29:55.650 00:29:58.889 Rico Rejoso: Alrighty, thank you so much for your time, guys. Talk to you.

370 00:29:59.130 00:30:00.150 Awaish Kumar: Thank you.

371 00:30:00.420 00:30:01.829 Samuel Roberts: Have a good one, guys. Bye.