Mar 31, 2026 Enterprise Brain Scrum - Transcript 00:00:00 Yeshwanth Reddy Yerraguntla: Hi, bro. Rajashekar G: Hi now. Yeshwanth Reddy Yerraguntla: What's up? Rajashekar G: Yeah. So yesterday we have deployed a version Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: uh right now a few visualizations likely design team still has to figure out and provide us the visualation components. Yeshwanth Reddy Yerraguntla: Why will design team do that? I thought uh it was on us to deliver Rajashekar G: Yes, Yeshwanth Reddy Yerraguntla: right. Rajashekar G: they will come up with the uh visualizations using D3 and animations related stuff. We are still figuring it out like Yeshwanth Reddy Yerraguntla: Are they figure are you saying they are figuring out what are the right uh uh right types of charts to show? Rajashekar G: yes. So right now uh day before yesterday we discussed with Navaru and we told him Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: we'll go with the basic uh whatever bar chart, pie chart, donut chart, line graph are there in our simulator P what Ganesha has shown we will be showing them we'll be using that package and Yeshwanth Reddy Yerraguntla: Okay. Okay. 00:02:41 Rajashekar G: uh representing them in our current uh system. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So Vara Kumar Jagarapu: Is that explicitly conveyed? trash shaker because like uh whatever we are Rajashekar G: which one Vara Kumar Jagarapu: showing right plotly charts right uh so lot of rendering everything and showing it once right but yeah there is no sync Rajashekar G: right Vara Kumar Jagarapu: up now yeah nar Rajashekar G: H no latter like no sync up has been happenedaru yesterday around 7 p.m. he came to me and told like uh we will sync up today for the review. Yeshwanth Reddy Yerraguntla: Today's the deadline, Vara Kumar Jagarapu: Okay. Yeshwanth Reddy Yerraguntla: right? Vara Kumar Jagarapu: Yeah. I'm not talking about I'm just because like day before yesterday when you Yeshwanth Reddy Yerraguntla: Got Rajashekar G: Yeah. Vara Kumar Jagarapu: uh mentioned that whatever the code that we are getting from the x is not up to the mark that we can use directly later they can vary to get into a structure right still they Rajashekar G: Yes. Vara Kumar Jagarapu: are working there might be an assumption apart from this whatever we are rendering right there might be expecting getting a few components from them as well. 00:04:11 Vara Kumar Jagarapu: Yeah, that is what running in my mind. But yeah, explicitly we need to mention. Rajashekar G: to mention. Vara Kumar Jagarapu: Yeah. Rajashekar G: Yeah. Vara Kumar Jagarapu: Anyway like if remaining things are done now extracting those components and we need to uh publish to the library and we need to import here right that we are Rajashekar G: Sure. Vara Kumar Jagarapu: working Rajashekar G: So this is Vara Kumar Jagarapu: as one more input one more input right so uh there are actually Rajashekar G: the Vara Kumar Jagarapu: five use cases uh we discussed right for one use case uh We gone through sheets and created tables and everything. Remaining things also in progress. Yeshwanth Reddy Yerraguntla: What? Vara Kumar Jagarapu: For that one sheet, we need to come up with some flows like uh from the demo perspective. Right? In that flow, Yeshwanth Reddy Yerraguntla: Okay. Vara Kumar Jagarapu: what are the visibles that are better to show them? In that way maybe we can prioritize and inform UX team so that they will give Yeshwanth Reddy Yerraguntla: Thank you. 00:05:18 Vara Kumar Jagarapu: those visuals first. I mean that is running in my back of my mind but yeah I from same to Rasha but you have some words maybe we can discuss. Oh, uh, do we have those flows are still working on Rajashekar G: So this questions like Prabakers are provided yesterday Vara Kumar Jagarapu: creating? Rajashekar G: night. This you need to see you need to check with the system like how they are providing the Vara Kumar Jagarapu: Okay. Rajashekar G: answers and I have written few questions by Yeshwanth Reddy Yerraguntla: Should be shot. Rajashekar G: using solve it. So this seems Vara Kumar Jagarapu: uh my interest here is uh I just want to know like what are the visuals that we need that Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: like Vara Kumar Jagarapu: we are going to show in that part of demo because I'm based on Rajashekar G: these are the these are Vara Kumar Jagarapu: so Rajashekar G: the Yes, Yeshwanth Reddy Yerraguntla: Where are they? Vara Kumar Jagarapu: anyway Yeshwanth Reddy Yerraguntla: Are you sharing? Rajashekar G: and I'm sharing my screen. 00:06:20 Yeshwanth Reddy Yerraguntla: Uh can you reshare? Vara Kumar Jagarapu: it is not visible Avenue Yeshwanth Reddy Yerraguntla: Maybe it's an issue from myself. Rajashekar G: Okay, sorry. Yeshwanth Reddy Yerraguntla: Okay, Vara Kumar Jagarapu: foundation. Yeshwanth Reddy Yerraguntla: I can see now. Rajashekar G: So these are the questions like uh uh solve it has provided if I ask it. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: So these are all like scenes five type of scenes are there. These are the actual questions we need to ask. This should be the output right here. Let's uh I'll try this thing like Yeshwanth Reddy Yerraguntla: I'm confused. What do you mean? Where is our money? What data set is it talking to? Rajashekar G: the one sheet like NSC this sheet early warning quotation Yeshwanth Reddy Yerraguntla: I see sorry. Rajashekar G: NC. Yeshwanth Reddy Yerraguntla: Yeah, I got some gap in the context. The total value of active contracts. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: So it ran an SQL query and generated this. Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: Okay. 00:07:31 Yeshwanth Reddy Yerraguntla: So from what I understand the only thing remaining is to show good Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: visualizations. Rajashekar G: good visualizations and uh a few peers have been generated uh uh user management, user login management and all the history we are connected with the Yeshwanth Reddy Yerraguntla: I see. Rajashekar G: database. Yeshwanth Reddy Yerraguntla: Got it. Makes Rajashekar G: So those things need to be merged. Yeshwanth Reddy Yerraguntla: sense. Rajashekar G: Uh I have given some feedback to Harshit on his PR and Ablash who have done streaming part like the streaming uh we are going with SS and right now not like Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: SS is working that also we Yeshwanth Reddy Yerraguntla: by that also we need Rajashekar G: need yeah those two things will be merged Yeshwanth Reddy Yerraguntla: to Rajashekar G: today before the Yeshwanth Reddy Yerraguntla: okay so what is the advantage of SSC over web Rajashekar G: de Yeshwanth Reddy Yerraguntla: sockets Rajashekar G: so SSC is very lightweight I And what Ganesh suggested is in optin they have already implemented. Yeshwanth Reddy Yerraguntla: Mhm. Rajashekar G: So websockets is getting complicated once these people are trying to implement. 00:08:34 Yeshwanth Reddy Yerraguntla: See okay but you are aware that websockets is superior because when you do some visualizations all those things back and forth if it's happening it can only happen with the websockets. Rajashekar G: Okay, sure. Yeshwanth Reddy Yerraguntla: Okay I one second I'll take my word back. Rajashekar G: I'll Yeshwanth Reddy Yerraguntla: All I am saying is if it can happen with SSC, I'm okay with that also. But there has to be back and forth communication. So if I select a specific bar in this uh this chart for example and then Rajashekar G: right. Yeshwanth Reddy Yerraguntla: if I ask uh give me the breakdown of this company it should know that I selected that and when I say this it means that particular bar something like yeah that is crucial Rajashekar G: Got it. I'll check that. Yeshwanth Reddy Yerraguntla: for experience center whatever we will deliver in M3 that is critical. Rajashekar G: Yeah, Yeshwanth Reddy Yerraguntla: Well, okay. Rajashekar G: sure. Yeshwanth Reddy Yerraguntla: One more uh one more I had is whatever merge request is happening like once you finish it, you have to sit and uh refactor the code so that uh these are reusable components. 00:09:53 Rajashekar G: Yes or no? We are doing it. Yeshwanth Reddy Yerraguntla: If if it's already that way, I'm very happy. Rajashekar G: We are doing it. Uh uh one concern I have like um a variable like front end structure is there right whatever we have given to cloud architecture. Vara Kumar Jagarapu: Yes. Rajashekar G: So there are I'm having few concerns let's discuss them like once we are in office. Vara Kumar Jagarapu: Yes. Rajashekar G: So right now this is the Vara Kumar Jagarapu: One more we uh I mean we are trying to take uh I mean create a package based Rajashekar G: system. Vara Kumar Jagarapu: on the uh internal uh chart package that we using. Yeshwanth Reddy Yerraguntla: for front end. Vara Kumar Jagarapu: I mean for example we are having this widget set all type of plot will be plotly visual. Yeshwanth Reddy Yerraguntla: Perfect. Yeah. Yeah. Perfect. Perfect. That's exactly what I want. Rajashekar G: These are coming from the experience center library Ganesh has deployed right that one that npm Vara Kumar Jagarapu: Yeah. 00:10:56 Yeshwanth Reddy Yerraguntla: Very good. Rajashekar G: package. Yeshwanth Reddy Yerraguntla: At the end of the day, we have to simply reuse all that library with minimal effort on editing the style, Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: right? uh like I was playing with something and added one feature in viasa it's not in prod but uh basically I click one button and it will change the whole Vara Kumar Jagarapu: I don't Yeshwanth Reddy Yerraguntla: theme so metallic background or something else something else something Vara Kumar Jagarapu: want Yeshwanth Reddy Yerraguntla: else it's like should become so trivial that when a new customer comes for them customizing should be basically just one CSS file file or something like that and everything in the UI should be controllable including the chart uh colors. So that's the goal. Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: If because you are meeting sorry we are recording this meeting you can capture this and put it in Jira backlog. Rajashekar G: Sure. So actually it is uh 70% that way Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: only the customization for the npm package will completely able 00:12:17 Yeshwanth Reddy Yerraguntla: All Rajashekar G: to so whatever the design system they have given we are using that colors Yeshwanth Reddy Yerraguntla: right. Rajashekar G: and injecting into the package so those are only rendering here that is what Yeshwanth Reddy Yerraguntla: Yes. Nice. Rajashekar G: we and the back end related Yeshwanth Reddy Yerraguntla: Yeah. close. You can just kill that uh window Vara Kumar Jagarapu: window itself. The small Yeshwanth Reddy Yerraguntla: this uh ex. Rajashekar G: Okay, Vara Kumar Jagarapu: window Rajashekar G: got it. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: So we have taken a branch chanika/ front end from chanika/alpha. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: We are mentoring everything to this branch. Right now once everything is stabilized here and ready all the refactor also done Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: we will move this to channel/ alpha. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: Uh so here I wanted to show you a few things. Vara Kumar Jagarapu: I hear Yeah. Rajashekar G: Actually cloud files are there around 50. I'll show it in 00:13:34 Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: local. Vara Kumar Jagarapu: Thank you. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So here in the back end structure there is chanika folder Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: right. And there is RT agent.py is there and RT agent is also Yeshwanth Reddy Yerraguntla: Correct. Rajashekar G: there. So we need to uh remove one of these and proceed with one thing actually whatever the changes we want to do for the core agent those are applying into the two Yeshwanth Reddy Yerraguntla: No, no, no. Rajashekar G: parts. Yeshwanth Reddy Yerraguntla: Sorry, I was talking on mute. You remove chakra.py because RT agent is just the refactored version of Chanaka. Ideally, Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: if you run if you remove chanaka.py and run make UI, it should work. Rajashekar G: Okay. Sure. Yeshwanth Reddy Yerraguntla: once that happens. Rajashekar G: We'll see like few changes has come into this. We'll discuss with the team and uh let it refactor. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: And then we have for the NL2SQL agent what we 00:14:50 Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: have done is I made it a core agent for NL2SQL in Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: which all the common things will be there for NLTSQL what needs to happen and insurance Yeshwanth Reddy Yerraguntla: Got it. Got it. Rajashekar G: DB and Tata still DB comes like these things can be plug and Yeshwanth Reddy Yerraguntla: Perfect. Rajashekar G: plug and that's why we have handling this will be exported in one port and this will be exported Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: in one port common things will be in this folder. Yeshwanth Reddy Yerraguntla: Nice. Rajashekar G: This is what the project is right now. For the front end, everything is in front end folder. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: There it Yeshwanth Reddy Yerraguntla: So, one question I have is if I run my AQI, Rajashekar G: is. Yeshwanth Reddy Yerraguntla: I'm assuming it will still launch pylog only. Correct? Rajashekar G: Yes, it will launch pylog only. Uh, Yeshwanth Reddy Yerraguntla: Uh Rajashekar G: and we need to run the front end to see the front end. 00:15:40 Rajashekar G: Uh, Yeshwanth Reddy Yerraguntla: got it got it. Basically it is launching pylo also but we are just not using it. Okay, Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: then you can continue using pylock for debugging. In case you Rajashekar G: Yeah. But response structure we need to uh see on like for pylog it Yeshwanth Reddy Yerraguntla: should Rajashekar G: is different structure for the current front end it is different structure. Yeshwanth Reddy Yerraguntla: actually yeah that I want to learn how you are actually making these two talk to each other. Vara Kumar Jagarapu: What the hell? Yeshwanth Reddy Yerraguntla: Now you said server side rendering, serverside events. Um but is there like a proper explanation of um what Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: uh port or what uh what is it called? Yeah. What connection is being established between front end and back end and uh like what is the connection route for asking a question like front end to back end what is the root back end to front end what is the root and where it 00:16:45 Rajashekar G: Yeah, previously like ask agent is there right? So here something is getting changed and it's not ask agent it is like ask Yeshwanth Reddy Yerraguntla: H. Rajashekar G: streaming that why it is streaming actually I need to Yeshwanth Reddy Yerraguntla: Mhm. Rajashekar G: see Yeshwanth Reddy Yerraguntla: Interesting. like why I'm asking is if I have to connect the front end to each port each agent individually like what does it take and basically the communication protocol how it is working I I want to learn Rajashekar G: I'll I'll ask for a document. I'll put a response document. I'm not clear right Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: now. Yeshwanth Reddy Yerraguntla: Yeah, you can just generate on markdown file and uh dump Rajashekar G: H so this is actually the end point query Yeshwanth Reddy Yerraguntla: it. Rajashekar G: SL so we will be using like based on the if it is a text contain it will continuously Yeshwanth Reddy Yerraguntla: Got it. Rajashekar G: streams will chunks the response like uh text will be continuously streamed and the data visualation part any JSON is coming once comes that will be pushed into the stream that is how it is happening 00:18:14 Yeshwanth Reddy Yerraguntla: Okay, got it. Fair enough. I think uh I don't need to ask those questions anymore. only yeah the my big concern is if I have to if I'm sparring a fresh agent how easy is it connect is it to connect to the UI that's the Rajashekar G: far away Yeshwanth Reddy Yerraguntla: only and if I want to have my every one UI for Rajashekar G: actually. Yeshwanth Reddy Yerraguntla: every agent what does it Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: take Rajashekar G: Uh this one the white base URL right now we are connecting with 8010. This is for deployment here. 8010 comes and six. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: The 8010 is our RT agent. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: That one only be exposed when we are deploying the Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: API. So we are talking with that if we want to have another UI Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: that will communicate with uh Tata steal or something directly we need to change this uh URL and deploy another version. 00:19:25 Yeshwanth Reddy Yerraguntla: Okay, Rajashekar G: That is how it happen like we are we can talk with 8010 or we can talk with another ports directly same like pyog but front end instances will be Yeshwanth Reddy Yerraguntla: got it. Rajashekar G: double like whenever you want to more communications like that we need more and one more concern I Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: have is right now for the AR orchestration purpose we have deployed RT agent data source agent. So user will be asking RT agent data source uh RT agent is asking the data source agent getting the answer and it is mod moving to here and it will be coming out as it is. This is having no responsibility of doing anything. It is only acting as a communication channel. Okay. Vara Kumar Jagarapu: Rasher meanwhile I just gone through like difference between the websocket and SSC, right? Rajashekar G: H. Vara Kumar Jagarapu: So the server sending almost like uh the events and the browser is catching those events, right? So it's a one one direction communication. Rajashekar G: Okay. Vara Kumar Jagarapu: Okay. 00:20:50 Vara Kumar Jagarapu: So the problems we see is sometimes right like for example uh the charts the dashboard in this like let's say we have score updates something like that we will be having for that use cases it is fine for example alerts notifications okay but like when there is some birectional communication so we have uh and one more thing is is not Rajashekar G: the interaction. Vara Kumar Jagarapu: properly standardized okay so keep in this mind like I think Rajashekar G: Okay, Vara Kumar Jagarapu: like we have to go for websockets or yeah Rajashekar G: that's all. Vara Kumar Jagarapu: do Rajashekar G: Sure. Vara Kumar Jagarapu: sorry you are able to hear me right Yeshwanth Reddy Yerraguntla: Yeah, I can hear you. Vara Kumar Jagarapu: no I just when the difference between SSC and this website okay but yeah they're not suggesting because it's not standard and is only perfect for alerts something like uh live dashboards and some charts. But yeah, uh there is no B birectional communication established throughout this. There will be HTTP connection but always server is going to send updates only browser is listening to them. 00:22:02 Vara Kumar Jagarapu: It's kind Yeshwanth Reddy Yerraguntla: Correct. Vara Kumar Jagarapu: of Yeshwanth Reddy Yerraguntla: There is H web sockets needed only during live uh dashboard updates experience center what um they call right that needs to Vara Kumar Jagarapu: Yeah. Yeshwanth Reddy Yerraguntla: be dynamic. Vara Kumar Jagarapu: Yes. Yeshwanth Reddy Yerraguntla: So I mean if I might be wrong and stupidly saying we should use Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: websockets when the time comes if you can if you clearly know that no no websockets is an Vara Kumar Jagarapu: Yeah. Yeshwanth Reddy Yerraguntla: overkill I will not I'm just making you remember that that requirement is there that's all I Rajashekar G: Yeah. Vara Kumar Jagarapu: Yeah. Yeah. So yeah, Yeshwanth Reddy Yerraguntla: give it to Vara Kumar Jagarapu: my intention also like we might use some other way but not SSC most Yeshwanth Reddy Yerraguntla: you Vara Kumar Jagarapu: probably that what I'm thinking. Yeshwanth Reddy Yerraguntla: sir Okay. Okay. You saying Rajashekar G: Yeah, Yeshwanth Reddy Yerraguntla: something? Rajashekar G: right now yesterday I discussed with V also yesterday. So we are comm user is communicating with RT agent. 00:22:58 Rajashekar G: RT agent is communicating with data source agent and data source agent doing all the processing and giving the information Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: to RT agent. RT agent directly providing that answer to user right now. Yeshwanth Reddy Yerraguntla: In the case of a single user? Rajashekar G: So we are yes we are having no responsibility here Yeshwanth Reddy Yerraguntla: Yes. Rajashekar G: nothing. Yeshwanth Reddy Yerraguntla: Nothing. Rajashekar G: So we are like uh right now uh to demonstrate the multi- aent orchestration we deployed it this way only. Here the kill is uh if we overuse it the tokens consumption will be uh more Yeshwanth Reddy Yerraguntla: You can explain it Rajashekar G: cover if an example every Yeshwanth Reddy Yerraguntla: example. Rajashekar G: question will pass through here and here and answer also pass through here and here. Yeshwanth Reddy Yerraguntla: H Rajashekar G: So this is doing nothing only standing in between but when but this LLM will consume the tokens. Yeshwanth Reddy Yerraguntla: fine fine then is your plan to not use is real time today Rajashekar G: We used it to demonstrate the data agent multi- 00:23:54 Yeshwanth Reddy Yerraguntla: bypass Rajashekar G: aent orchestration. If not needed after the demo, we will only communicate with this later that the Yeshwanth Reddy Yerraguntla: you know for for the sake of demo Oh Rajashekar G: data. Yeshwanth Reddy Yerraguntla: see wait basic question in the demo you have multiple agents right multiple data sources right okay then you'll always need real time agent Rajashekar G: Yes. Uh yes, Yeshwanth Reddy Yerraguntla: no Rajashekar G: but we only dealing with Tastel right now. So remaining we not need an my point which Yeshwanth Reddy Yerraguntla: okay that's what I was asking so one source agent data source Okay, Rajashekar G: deploy. Yeshwanth Reddy Yerraguntla: then drop it off. For the sake of demo, it's okay. Rajashekar G: Okay. Once done we will make those Yeshwanth Reddy Yerraguntla: So, Rajashekar G: changes. Yeshwanth Reddy Yerraguntla: I have a a different approach for this token duplication checker is a very important concept especially when you're dealing with Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: um uh data sources which will give you tons and tons of data. For example, 00:25:01 Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: by mistake, if it is returning something like thousand rows, I should the data source agent should not first of all Rajashekar G: Mhm. Yeshwanth Reddy Yerraguntla: if it is going to send the,000 rows as out output to realtime agent. It has to write those thousand rows. First of all, it is reading those thousand then it is writing those thousand. So, you're wasting 2,000 tokens. Now, Rajashekar G: Right. Yeshwanth Reddy Yerraguntla: all you have to do is something called pass by reference. You create an intermediate file. That file you pass it to real time data say agent saying I wrote the result here. Rajashekar G: H. Yeshwanth Reddy Yerraguntla: You go and check it. That way um uh you are that is how we also as Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: a human beings we also do the same thing right when I have some some major thing to discuss with you ideally I should not wait till the Google meet call and then I talk everything that is in my mind ideally I should write it down somewhere give you the link you browse it at your leisure 00:26:00 Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: you consume it however you want to consume we should do the same thing for all data source Rajashekar G: Yeah, Yeshwanth Reddy Yerraguntla: agents Especially where there is large uh uh input output Rajashekar G: got it. Yeshwanth Reddy Yerraguntla: exchange. Rajashekar G: Shut Yeshwanth Reddy Yerraguntla: Prairana you is there Rajashekar G: up. Satyasri Prabhakar Mantripragada: I just got to understand. Yeshwanth Reddy Yerraguntla: anything? Satyasri Prabhakar Mantripragada: Did somebody tell you that Tata still is a demo? Rajashekar G: Pattern sir demo. Satyasri Prabhakar Mantripragada: We may have to deploy Rajashekar G: Uh Satyasri Prabhakar Mantripragada: because if this needs to be deployed onto them Rajashekar G: So actually today today will be a demo demo Satyasri Prabhakar Mantripragada: internal hands assume that if we have to hand over Rajashekar G: H. Satyasri Prabhakar Mantripragada: a feel of it and we should be able to deploy it also on then do you have to make any other det. Rajashekar G: Okay. Satyasri Prabhakar Mantripragada: changes something that you are discussing demo work Rajashekar G: Mhm. Satyasri Prabhakar Mantripragada: is ready for deployment develop Rajashekar G: Right. Satyasri Prabhakar Mantripragada: direction. 00:28:07 Satyasri Prabhakar Mantripragada: Are we doing that? Rajashekar G: Uh this is ready for deployment and whatever is getting done that is ready for deployment but my only concern is uh changes and based on the discussions few fixes production deployment but my concern Satyasri Prabhakar Mantripragada: Okay. Oh, that's Rajashekar G: is our milestone for Wednesday is to demonstrate the multi- Satyasri Prabhakar Mantripragada: fine. Rajashekar G: aent orchestration and the reactive application on Tata data. So R fulfill once done Satyasri Prabhakar Mantripragada: Okay, fine. Rajashekar G: we can take a call and uh uh down other things whatever not needed and only deploy what needed Yeshwanth Reddy Yerraguntla: checker. Satyasri Prabhakar Mantripragada: Okay. Yeshwanth Reddy Yerraguntla: You should also prepare a deployment u diagram how things are being Rajashekar G: yeah pipelines are there yesterday I sit with Nitish you Yeshwanth Reddy Yerraguntla: deployed. Rajashekar G: only running uh uh and uh same enterprise brain server low he is deploying it like uh same way however our docker docker file we have written he's only running the makea command and that is getting deployed only he's exposing the endpoint what we needed that is the back end 00:29:42 Yeshwanth Reddy Yerraguntla: Why I'm asking is yesterday we had a call with one of the clients opt in. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: What they basically asked is you we gave them the deployment uh one month ago. They have been using it. Rajashekar G: Mhm. Yeshwanth Reddy Yerraguntla: Yesterday they came back and asked uh we want to now take it completely from you and uh start working on it ourselves. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: And one of the question they asked was how have you currently deployed? One mistake we like we sort of um whether we completely understood their question or not is I will not debate that but the mistake from their perspective what we have delivered as a diagram was the ideal diagram this is how you should deploy K we gave a Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: picture they came back and said we will it's always Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: easy to come up with aspirational diagram s how did you currently deploy? Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: So in that deployment we kept on answering one question after the other one and just to answer all those questions it took almost whatever 20 30 minutes um what do you call where is the 00:30:58 Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: logging is it in all the questions they kept on asking. So whenever you deploy something immediately if you can bring generate a deployment snapshot what are commands to bring it up what where is each and everything sitting how is each module talking to the outside world what are the environment variables everything if you can generate that Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: one deployment this thing um it'll be very good on your part that you know you are taking care of all these Rajashekar G: Sure enough. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: All right. Rajashekar G: Got Yeshwanth Reddy Yerraguntla: So, Rajashekar G: it. Yeshwanth Reddy Yerraguntla: okay, I'll have to drop for the other call. Rajashekar G: Yeah. Uh and the next thing in the sprint we are having the training life cycle. So after this milestone by end of the end of the sprint we need to provide the training life cycle. So once you are free or uh tomorrow morning we will think about like how that also need to be implemented. 00:32:23 Yeshwanth Reddy Yerraguntla: Come up with the plan. Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: Then we'll discuss what are the open questions, Rajashekar G: Sure. Yeshwanth Reddy Yerraguntla: what is the scope, what is the expected behavior. Rajashekar G: Yeah. And Yeshwanth Reddy Yerraguntla: Okay. Satyasri Prabhakar Mantripragada: Not related to this. Rajashekar G: yesterday Yeshwanth Reddy Yerraguntla: The I saw the three use Satyasri Prabhakar Mantripragada: Did you get a chance? Yeshwanth Reddy Yerraguntla: cases of the three. First one is super feasible, super valuable. Satyasri Prabhakar Mantripragada: Uh uh but they thinking that okay second one is what is Yeshwanth Reddy Yerraguntla: Second one is but second one is like first of all deep learning related and I don't think it even has medical this thing Satyasri Prabhakar Mantripragada: referring. Yeshwanth Reddy Yerraguntla: like I like I person because I worked in medical domain I got very I don't know like there were so many red flags in that uh aspect that it's not possible because of a lot of reasons we can discuss it afterwards first and third feasible in my head at least from what Satyasri Prabhakar Mantripragada: Uh 00:33:28 Yeshwanth Reddy Yerraguntla: I read and what I must try to analyze. Satyasri Prabhakar Mantripragada: firstly we identified Yeshwanth Reddy Yerraguntla: True. Satyasri Prabhakar Mantripragada: existing voice we can get it Yeshwanth Reddy Yerraguntla: Yes. Satyasri Prabhakar Mantripragada: connected. Rajashekar G: Good s***. Yeshwanth Reddy Yerraguntla: Yes. Satyasri Prabhakar Mantripragada: So it is more handheld easy. Yeshwanth Reddy Yerraguntla: Yes. Satyasri Prabhakar Mantripragada: So that is that is what we have thought. But the other thing that we wanted to show is like how how quickly can you save a patient. You should know what is the current status current situation based on parameters or last one hour parameters. If something get syncp three parameters if anyities are observed you can direct that to okay this kind of a symptom is getting the medical staff who is attending the patient may not be immediately aware only the senior physicians somebody may be an expert in diagnosing that kind of thing at the earliest. Yeshwanth Reddy Yerraguntla: Okay. In that case, the wording was wrong. Satyasri Prabhakar Mantripragada: So I learn to uh the 00:34:55 Yeshwanth Reddy Yerraguntla: Yeah. Okay. Sure. That can happen. Satyasri Prabhakar Mantripragada: objective is that Yeshwanth Reddy Yerraguntla: I think we can try it. It's a time series analysis. Satyasri Prabhakar Mantripragada: uh sample monitor but what Yeshwanth Reddy Yerraguntla: Okay. Satyasri Prabhakar Mantripragada: we are asking them is give us 10 15 such patients information for over a period of one year at least for this kind of a case it is only cardiology related. So if we can get the data of Yeshwanth Reddy Yerraguntla: My question is how can a patient sorry one question I have is how can a patient be Satyasri Prabhakar Mantripragada: that Yeshwanth Reddy Yerraguntla: in ICU for one year? Satyasri Prabhakar Mantripragada: not for one year doctor one year different different cases this is one patient in an ICU at this moment okay as long as that patient is there we'll collect his ICU Yeshwanth Reddy Yerraguntla: H. Satyasri Prabhakar Mantripragada: data and two days from the 3 days on the patient Okay. Each patient will have an ID. So there may be recurring cases also. 00:36:10 Satyasri Prabhakar Mantripragada: over a period of one year duration and within an ICU if there are 10 beds we are asking 10 such patients data over a period of one year different kinds of problems we'll discuss it later but that is the high level goal Yeshwanth Reddy Yerraguntla: Yeah. Yep. Now like that one sentence probably was carrying too much information. Now that you explained I can rethink. Yeah. Satyasri Prabhakar Mantripragada: It was suggested that within monitor Yeshwanth Reddy Yerraguntla: H. Satyasri Prabhakar Mantripragada: I'll have to prepare. I have it ready. I got itated through Yeshwanth Reddy Yerraguntla: Okay. Satyasri Prabhakar Mantripragada: some then we can start. Meanwhile, Yeshwanth Reddy Yerraguntla: Yeah. Satyasri Prabhakar Mantripragada: we'll close the startup field and then one team has to work on the core platform because some other projects will keep coming. We'll have to deliver those and we'll have to evolve the platform Yeshwanth Reddy Yerraguntla: Yeah. Satyasri Prabhakar Mantripragada: also. Yeshwanth Reddy Yerraguntla: Whatever checker is building the UI, that's the one that we'll constantly reuse. Satyasri Prabhakar Mantripragada: Fine. Fine. Yeah. Thank you. Yeah. Thank Rajashekar G: Okay, sir. Thank Satyasri Prabhakar Mantripragada: you. Rajashekar G: you. Yeah. Satyasri Prabhakar Mantripragada: If that matches to that answer then we are Rajashekar G: Yeah. Okay. Satyasri Prabhakar Mantripragada: good. Rajashekar G: Sure. It's like NLSQL whatever we have given previously my insurance database allowancer Satyasri Prabhakar Mantripragada: uh technical if I have to test them some Q needs to test automatic way are there any uh are there any Rajashekar G: right now manual right no it's not there my AP calls Satyasri Prabhakar Mantripragada: API calls being done uh Rajashekar G: can be done Satyasri Prabhakar Mantripragada: but can be built uh Rajashekar G: apost tool. We can hit those APIs and get the answers and evaluate. Satyasri Prabhakar Mantripragada: Okay. Automation. Rajashekar G: Yes sir. Transcription ended after 00:40:20 This editable transcript was computer generated and might contain errors. People can also change the text after it was created.