Meeting Title: Uttam Kumaran Date: 2025-03-20 Meeting participants: Casie Aviles, Uttam Kumaran, Amber Lin


WEBVTT

1 00:02:21.300 00:02:22.450 Uttam Kumaran: Hey, Casey?

2 00:02:26.200 00:02:26.820 Casie Aviles: Hey!

3 00:02:29.790 00:02:30.720 Uttam Kumaran: Late.

4 00:02:32.130 00:02:35.199 Casie Aviles: Yeah, yeah, my, my brain’s not working anymore.

5 00:02:35.490 00:02:37.609 Uttam Kumaran: Okay. Okay, hey? I don’t want to keep you up

6 00:02:40.277 00:02:46.769 Casie Aviles: But but yeah, I mean, I I wanted to finish the dashboard stuff already, but I’m guessing

7 00:02:46.770 00:02:52.539 Uttam Kumaran: Let me tell me where we left off. I’ll take it. I’ll finish it up, and I’ll I’ll have it done by later today.

8 00:02:54.050 00:02:59.949 Casie Aviles: It’s just, you know, with the joins like I couldn’t get like

9 00:03:00.050 00:03:02.799 Casie Aviles: I don’t know. I couldn’t get it properly working

10 00:03:04.450 00:03:05.110 Uttam Kumaran: Okay.

11 00:03:05.912 00:03:10.630 Uttam Kumaran: do you have anything I can start from? Or I’m just gonna start sort of from Rpr. Yesterday.

12 00:03:12.270 00:03:20.190 Casie Aviles: Yeah, like with the I. I don’t think there’s like a way to link all of the

13 00:03:20.390 00:03:24.659 Casie Aviles: tables. I was hoping that it would be the output.

14 00:03:25.900 00:03:31.819 Casie Aviles: Yeah, the the part with brain trust where we don’t have the user input and the AI output. It’s

15 00:03:32.260 00:03:34.290 Casie Aviles: kind of sucks because I couldn’t.

16 00:03:34.650 00:03:38.610 Casie Aviles: We couldn’t have all everything in one table

17 00:03:43.430 00:03:46.639 Uttam Kumaran: Let me log into brain. Trust they have an Api right

18 00:03:48.940 00:03:52.890 Casie Aviles: Hmm, yeah, I think so. I was playing around with.

19 00:03:53.840 00:03:55.860 Casie Aviles: If there’s they have this query.

20 00:03:59.214 00:04:03.159 Casie Aviles: Yeah, I could share the link

21 00:04:17.260 00:04:18.200 Uttam Kumaran: Hmm!

22 00:04:18.810 00:04:20.109 Uttam Kumaran: What the heck is this

23 00:04:22.410 00:04:28.430 Casie Aviles: I’m trying to like, get the input output and expected. And also the scores

24 00:04:29.784 00:04:30.380 Uttam Kumaran: Okay.

25 00:04:30.910 00:04:33.229 Casie Aviles: They? I think they have this query.

26 00:04:33.460 00:04:36.270 Casie Aviles: yeah, brain test? Query language. So

27 00:04:37.060 00:04:38.729 Uttam Kumaran: But what happened here? It didn’t.

28 00:04:39.750 00:04:40.730 Uttam Kumaran: Oh.

29 00:04:45.653 00:04:48.129 Casie Aviles: I’ll I’ll send a screenshot of how?

30 00:04:58.440 00:05:02.200 Casie Aviles: Yeah. Because, for example, this image that I sent

31 00:05:02.340 00:05:07.439 Casie Aviles: it’s just this course. I don’t have the input output and expected

32 00:05:07.690 00:05:08.590 Uttam Kumaran: Hmm.

33 00:05:14.820 00:05:17.720 Casie Aviles: But when when I go into each row

34 00:05:18.100 00:05:22.259 Casie Aviles: I could get I yeah, I I would see the

35 00:05:29.630 00:05:31.879 Uttam Kumaran: So this. So this is an experiment.

36 00:05:33.800 00:05:34.590 Casie Aviles: Yeah.

37 00:05:35.290 00:05:39.420 Uttam Kumaran: So, input.

38 00:05:43.630 00:05:45.190 Uttam Kumaran: oh, this is the output

39 00:05:46.010 00:05:48.879 Casie Aviles: Yeah, I’m trying to get those and also the scores.

40 00:05:54.070 00:05:56.900 Casie Aviles: But they have to. Yeah, to select all of that

41 00:06:11.760 00:06:13.850 Uttam Kumaran: How can I select all this shit?

42 00:08:14.030 00:08:15.150 Uttam Kumaran: Hmm.

43 00:08:36.210 00:08:39.139 Uttam Kumaran: I mean, did you try this like just doing a list experiment?

44 00:08:41.315 00:08:41.799 Casie Aviles: No

45 00:08:41.980 00:08:46.490 Uttam Kumaran: Okay, okay, that’s fine. Then I’m gonna I’ll probably try to just run this and get a version of this

46 00:08:47.610 00:08:48.000 Casie Aviles: Okay.

47 00:08:48.000 00:08:51.039 Uttam Kumaran: I can just load it in. I’ll get all experiments, all

48 00:08:52.010 00:08:55.999 Uttam Kumaran: emails and stuff. I’ll just throw it into Snowflake, and then I’ll

49 00:08:57.070 00:09:03.180 Uttam Kumaran: I’ll let you know. But let me. I want to push your I want to push your

50 00:09:03.510 00:09:05.319 Uttam Kumaran: real VR.

51 00:09:06.382 00:09:09.409 Uttam Kumaran: Oh, shit! I think I may still have it open?

52 00:09:24.290 00:09:26.640 Uttam Kumaran: Is this good like? Are we good to go on this

53 00:09:33.400 00:09:36.189 Casie Aviles: I didn’t. I did not commit anything, though

54 00:09:37.680 00:09:38.920 Uttam Kumaran: Do you have anything new?

55 00:09:39.580 00:09:41.070 Casie Aviles: No, no, not at all.

56 00:09:41.260 00:09:42.660 Uttam Kumaran: Okay, okay, don’t worry about it, though.

57 00:09:43.430 00:09:44.243 Uttam Kumaran: Let me.

58 00:09:47.990 00:09:52.590 Uttam Kumaran: One thing I’m gonna do, though, before I do this is I want to make sure

59 00:09:52.770 00:09:57.529 Uttam Kumaran: like that. Everything here is is comes as 0 instead of null.

60 00:09:57.720 00:10:03.190 Uttam Kumaran: So what I’m gonna do is I will go into here.

61 00:10:05.870 00:10:08.698 Uttam Kumaran: Yeah. My brain is fried, too.

62 00:10:10.100 00:10:12.400 Uttam Kumaran: What the heck am I looking at?

63 00:10:20.060 00:10:22.910 Uttam Kumaran: Oh, okay.

64 00:10:26.290 00:10:29.709 Uttam Kumaran: what is this? Total conversations?

65 00:10:37.220 00:10:37.900 Uttam Kumaran: Hmm.

66 00:10:59.040 00:11:00.240 Uttam Kumaran: What’s there?

67 00:12:50.540 00:12:58.869 Uttam Kumaran: Oh, there’s literally nothing for that, timestamp. So it’s not even this that I’m doing.

68 00:13:41.280 00:13:48.439 Uttam Kumaran: So I need to have no conversations on those days

69 00:13:49.240 00:13:52.010 Uttam Kumaran: but a record for those days.

70 00:13:53.000 00:13:57.599 Uttam Kumaran: So what the heck. How do I do that?

71 00:15:41.160 00:15:43.080 Uttam Kumaran: Hmm!

72 00:15:43.720 00:15:49.680 Uttam Kumaran: I don’t see any like know this stuff out if it’s not working. So

73 00:16:27.880 00:16:28.690 Uttam Kumaran: hmm!

74 00:16:29.630 00:16:37.350 Uttam Kumaran: There’s an interesting problem. So, Casey, you’re following. So what’s happening is this, data doesn’t have any.

75 00:16:37.480 00:16:46.539 Uttam Kumaran: There’s no data for the execution times that we need like the ones in the weekends. What’s happening is

76 00:16:46.910 00:16:55.900 Uttam Kumaran: the dashboard is running on Event timestamp.

77 00:16:56.480 00:17:03.560 Uttam Kumaran: But there is no data for that event, Timestamp. So it’s sort of just puts it at 0.

78 00:17:03.840 00:17:08.119 Uttam Kumaran: Guess what I’m trying to see is like how I can avoid that.

79 00:17:25.140 00:17:26.480 Uttam Kumaran: I guess.

80 00:17:32.830 00:17:33.670 Uttam Kumaran: Hmm!

81 00:17:35.870 00:17:38.500 Uttam Kumaran: I guess what I could do is

82 00:17:42.110 00:17:48.310 Uttam Kumaran: put this all no.

83 00:17:52.740 00:17:53.490 Uttam Kumaran: Hmm.

84 00:17:54.550 00:17:59.300 Uttam Kumaran: How do I prevent it from showing those days?

85 00:18:30.010 00:18:34.629 Uttam Kumaran: Okay, maybe let’s just ask. Let’s just ask AI what it thinks we should do.

86 00:18:39.720 00:18:41.669 Uttam Kumaran: what is this? So

87 00:18:55.850 00:19:01.680 Uttam Kumaran: we don’t have any data for the weekend. And currently, it’s coming in as

88 00:19:01.800 00:19:04.909 Uttam Kumaran: null. When I just wanted to come in as 0.

89 00:19:05.160 00:19:08.640 Uttam Kumaran: What changes should I make to the real model?

90 00:19:53.410 00:19:56.259 Uttam Kumaran: I don’t know if this guy’s gonna be able to help. Here.

91 00:19:59.770 00:20:03.149 Uttam Kumaran: let’s keep let me look one more time at something

92 00:20:07.520 00:20:10.380 Casie Aviles: We’re we’re trying to remove those blank parts right?

93 00:20:10.380 00:20:12.499 Uttam Kumaran: I just want it to come in as 0

94 00:20:12.900 00:20:13.820 Casie Aviles: Oh, okay.

95 00:20:29.920 00:20:32.879 Uttam Kumaran: This is so bad that this guy sucks.

96 00:20:33.420 00:20:34.413 Uttam Kumaran: Okay, let’s

97 00:20:40.690 00:20:48.370 Uttam Kumaran: let me try one more thing. So if I is there a default, I have to choose a default.

98 00:20:48.930 00:20:54.580 Uttam Kumaran: If let’s just do this, and the expression is

99 00:20:58.280 00:21:00.709 Uttam Kumaran: time series is event time. Stan.

100 00:21:02.670 00:21:05.050 Uttam Kumaran: I can just say.

101 00:21:10.880 00:21:16.970 Uttam Kumaran: I try that and nothing.

102 00:21:20.080 00:21:21.160 Uttam Kumaran: Hmm.

103 00:21:42.090 00:21:45.709 Uttam Kumaran: okay, maybe just not worth

104 00:21:56.680 00:22:00.219 Uttam Kumaran: not worth figuring this out now. But

105 00:22:13.370 00:22:17.999 Uttam Kumaran: okay, whatever let’s, I’m just gonna not deal with this right now.

106 00:22:18.880 00:22:24.900 Uttam Kumaran: So great. So we have. Let’s can you tell me, Casey, what’s the reason? The record?

107 00:22:25.380 00:22:27.039 Uttam Kumaran: And then what is a?

108 00:22:27.610 00:22:31.280 Uttam Kumaran: Oh, we have okay, conversation, sessions and records.

109 00:22:32.150 00:22:34.489 Uttam Kumaran: So what’s the difference between these 2

110 00:22:38.570 00:22:39.630 Casie Aviles: Sessions.

111 00:22:40.340 00:22:44.179 Uttam Kumaran: Like what is conver? What is difference between conversations and sessions?

112 00:22:47.380 00:22:54.290 Casie Aviles: Let me know. I think it’s just with the session. Ids, I think that’s where I just asked AI to generate this

113 00:22:55.510 00:22:58.179 Uttam Kumaran: But like what is session? What is a session

114 00:23:15.480 00:23:22.310 Amber Lin: I was wondering about that, too. But I suppose because the conversation never really ends. Maybe it’s just

115 00:23:22.630 00:23:29.500 Amber Lin: a user started a session, and then they can have multiple conversations in the session. That’s my understanding.

116 00:23:29.500 00:23:34.960 Casie Aviles: I think it’s the unique sessions. Yeah, unique sessions recorded in the logs

117 00:23:35.360 00:23:40.020 Uttam Kumaran: Oh, but it’s but probably conversations and records is probably the better one. Right

118 00:23:41.610 00:23:45.230 Amber Lin: Probably I don’t think the client needs the sessions

119 00:23:46.020 00:23:50.150 Uttam Kumaran: Yeah, so records is like the back and forth conversation

120 00:23:51.720 00:23:52.290 Amber Lin: Isabel.

121 00:23:52.290 00:23:52.900 Uttam Kumaran: So.

122 00:23:52.900 00:23:53.929 Amber Lin: A topic.

123 00:23:55.320 00:23:57.430 Uttam Kumaran: Okay, so.

124 00:24:13.120 00:24:19.209 Uttam Kumaran: but a conversation is like, what is the between a conversation and a record

125 00:24:23.960 00:24:30.160 Uttam Kumaran: like our record is, yeah, what is the difference between these

126 00:24:44.230 00:24:46.490 Casie Aviles: The records have, like

127 00:24:47.050 00:24:49.619 Uttam Kumaran: Oh, there’s not even a session thing in here

128 00:24:52.490 00:24:53.310 Casie Aviles: Yeah.

129 00:24:53.990 00:24:55.700 Uttam Kumaran: But where’s session coming from?

130 00:25:03.170 00:25:04.909 Uttam Kumaran: What the heck wait, hold on!

131 00:25:13.200 00:25:19.950 Uttam Kumaran: Oh, so conversations will.

132 00:25:20.880 00:25:28.660 Uttam Kumaran: What conversations have session in them?

133 00:25:30.170 00:25:33.469 Uttam Kumaran: So conversations and sessions are basically the same thing.

134 00:25:35.750 00:25:38.510 Uttam Kumaran: So let’s just go ahead and use one of them.

135 00:25:39.720 00:25:42.120 Uttam Kumaran: So I’m gonna delete this.

136 00:25:43.430 00:25:49.209 Uttam Kumaran: And I assume that. Yes, records are one back and forth

137 00:25:49.590 00:25:58.850 Uttam Kumaran: cool. So I’m gonna change this, not to total records, but to total response pairs, maybe.

138 00:26:00.610 00:26:02.919 Uttam Kumaran: and that that’ll be fine.

139 00:26:06.210 00:26:08.150 Uttam Kumaran: So we have 264,

140 00:26:09.120 00:26:12.020 Uttam Kumaran: 679 response pairs.

141 00:26:12.500 00:26:14.049 Uttam Kumaran: What else can you give me?

142 00:26:16.910 00:26:22.240 Uttam Kumaran: Name is total response.

143 00:26:23.660 00:26:24.880 Uttam Kumaran: Pairs.

144 00:26:26.950 00:26:29.330 Uttam Kumaran: Conversation measure great.

145 00:26:30.160 00:26:34.239 Uttam Kumaran: So yep, around 2 and a half back and forth

146 00:26:36.240 00:26:40.189 Uttam Kumaran: between average execution. Time is 3 seconds

147 00:26:52.930 00:27:01.160 Uttam Kumaran: And then one thing I I think I’ll I have to jump right now for a sec. But one thing I’m gonna work on is probably a canvas dashboard for them as well.

148 00:27:03.930 00:27:09.240 Uttam Kumaran: But I feel pretty good about how this looks. The only thing is they don’t. I don’t. I think I’m just gonna put the

149 00:27:12.540 00:27:16.249 Uttam Kumaran: I’m gonna put the input output higher

150 00:27:17.120 00:27:22.480 Uttam Kumaran: because I don’t think they really are going to care about the record id much.

151 00:28:12.460 00:28:14.450 Uttam Kumaran: and maybe I can’t do that.

152 00:28:18.760 00:28:25.780 Uttam Kumaran: Oh, but this is the dashboard. So the da so in the so I can keep all these in the dashboard. I can just bring the one site. Okay.

153 00:28:26.750 00:28:31.100 Uttam Kumaran: the quality score. This one we’re gonna have to replace with the other score.

154 00:28:34.020 00:28:34.930 Uttam Kumaran: Okay.

155 00:28:38.160 00:28:41.040 Uttam Kumaran: and what’s the feedback model look like?

156 00:28:47.420 00:28:48.760 Uttam Kumaran: Great. Okay.

157 00:28:51.060 00:28:58.340 Uttam Kumaran: So, Casey, how do I join? How do I join this to to the messages

158 00:29:00.400 00:29:03.229 Casie Aviles: Think maybe through the AI output, but

159 00:29:05.920 00:29:10.889 Uttam Kumaran: Oh, I see. Okay. Okay. Okay.

160 00:29:11.220 00:29:11.980 Uttam Kumaran: Okay.

161 00:29:12.280 00:29:18.410 Uttam Kumaran: Alright dude. Get some. Get some rest. I’m gonna come back, Amber. I’ll probably I’ll be back in probably an hour, 30 min.

162 00:29:18.840 00:29:19.450 Uttam Kumaran: We could check

163 00:29:19.730 00:29:24.159 Amber Lin: Okay, get some rest. My brain is so fried I imagine yours, too.

164 00:29:25.960 00:29:27.670 Amber Lin: Alright, guys, I’ll talk soon.

165 00:29:27.670 00:29:28.500 Uttam Kumaran: Thank you.