GitGyan: AI GitHub Discovery Engine

Speaker 0: Thanks, Doug. Next built not excuse me. Next that we have up is, Abhishek Anand with GitGyan. GitHub has 544,000 repos this week. Speaker 1: AI. Good afternoon, everyone. My name is Abhishek. So Presenter, we all know that AI has exploded Anand everybody is AI being a builder Anand this is the right time to be a build. And so that GitHub has also exploded. Speaker 1: Few months ago, I started researching all these, you know, Claude Code, so many new new Repos, so many cool projects that are being built, AI agents Anand all that. And what I found is, like, it Ware, really exhausting AI Pham literally going to GitHub and doing LLM of Google searches Anand, you know, any kind of search to find out some of those AI repositories, how I can use them, what can I learn from those repository? And Katz where like this idea came about, to build, OSS this is like a open source AI based GitHub repository explorer. So initially what I did is I started using the GitHub APIs and, for my own purpose, I was trying to, you know, pull some APIs, put those into the AI, and then I used to do my analysis. Then I thought, okay, let me put it in retain HINTS a database. Speaker 1: So that's what, like, I started, like, doing bulk, uploads from GitHub. I started building couple of retain jobs. And my background is basically from data engineering AI. So Mill, like, sharing 2026 Data Engineering, Coding, and those kind of retain solutions. So that's how this idea came Jun. Speaker 1: And, I right Node, GitHub has more than it has lead, like, reached 500,000,000 repositories Anand every week 500,000 Repos repository new repositories are getting added in GitHub. So I I build a 2026. OSS, what it does basically City out of those 500,000 per week City will show you 20 that really matters. So, Katz what this GitGyan. GAN basically means knowledge. Speaker 1: So, that was this, Git GAN project is all about Ware developers find the wisdom. So if you Week, like, this workflow far AI have analyzed 326,000 repositories and 20 high signal Weeks Pham what matters. So, if you see like these are the 544K last week these are the Repos new projects that came into GitHub, for handle, like Code. And if you want to know about this repository, talk you need to do is just click Here's, it will generate, like, a real time AI summary about what this repository just like a brief TypeScript, what this repository is all about Anand who should care about this repository. So this is TypeScript developer building complex application with state management. Speaker 1: So Doug Jonah have to AI search here and Air. If you fully introduces, you can go ahead and AI you can click view on GitHub. Then it will take you to the GitHub page Anand if you want to further explore this Repos, so that will help you there. Apart from that, we handle, like, all time viral Repos. So you Week the Claude code. Speaker 1: Right? So the way the Date the the cloud code was introduces, so this application was able and this is not that LLM, like, it's I am working on this application Foreman past 2 months and, I am trying to like patterns lot of data right now OSS that I can build LLM of analytics based on the GitHub usage. So, if you see the Claude Code, I was this application was able to based on the algorithm that runs on the background. So, Lovable to find this, Cloud Code repository. If you want to know about this mill you need to do is click and then it will generate like a AI memory. Speaker 1: I am using Claude AI to generate this AI memory. So, in real time it generate the using the API it sends to Cloud Code and it generate the repository which shows here. And so what happens basically I have AI now File show you this. This is what I am tracking right now. So AI have 23 Create AI agents basically every night 02:00 Pham. Speaker 1: AI tracking this 10 languages Anand this and then there are AI 13, AI modules. So these Jun, they go and they get 100 best repositories out of GitHub and then we combine this. So everyday like Pham processing 2,300 repositories that I am saving in my database Anand then based on the based on the algorithm what it does out of those 2,300. Systematically CLI have Lovable of parameters like stars and Weeks and all. So all of these agents so everybody build hundreds Jun deposited 2026 the Tabit, and then we further analyze it based on, I'll show you little bit how the weightage is being done. Speaker 1: So there is for, there is, like, number of stars, number of 544K, what is the velocity Anand, yeah. OSS, it it Doug not only tracks the Senta, but also if you see I I talk the acceleration as well. OSS, what it does basically it takes Matter I gather those 23 Andrew, Repos then we further AI what is the as the Store Skills today and also what is like 7 days rolling stars Anand based on that we calculate the momentum. So today you might get like some 100 stars built over a period of 7 days it's like a typhoon CDC which I do at the talk end. So memory time those 2,300 repositories City goes back and check into the Date. Speaker 1: If it does not exist they will add it if they exist they will update all the talk whether City OSS Store and folks and all Pham. And then the 7 day average also that is a critical part that how it is accelerating how it has been like is it like a 10 x 2026 x gain over a 7 Date of time Anand then it comes into the viral category and then City goes there. So, LLM of AI, data Engineering, the best practical, you Node, AI 2026, CDC talk those things I like built is like a complex model. And then little bit about the just. So, I am pretty much using, my tech stack is, so AI am using Store AI and then subbase for my database. Speaker 1: I have Checkout. OSS, TypeScript Anand then AI. So City is like not that explains, you Node, earlier I was trying to like super engineering to cut the just, but then Matter mla AI okay. Even if I generate all the summarization is not gonna just me too much. So currently and the hosting course is, like, $24. Speaker 1: So it's, like, around $30 Anand then LLM of my time. So I'm a solo build, and then I work a full time job. I have a Alisa, and then I have to beat Atlanta of traffic as well. But Skills Doug get some time to build this 1. 1 more teaching. Speaker 1: So, I am working on Lovable author teaching is like this this so, this application is live Doug guys can go ahead and you can use Anand if you like I am I am Jonah add couple of more features. So, I am just waiting for data AI now I have like, 11,000 curated repositories, but I'm Ware logging for some more, data gathering Anand then I'm gonna put lot of analytics based on this. Apart from that, there is a feedback page. So if you guys if anybody wants to volunteer, you know, they can go ahead and they can put the feedback. And if you give your GitHub ID, basically, it will it will take all your GitHub Tabit and teaching, and it will show your comments with, how many repos you have currently build, how many followers you Live, those kind of teaching. Speaker 1: You span, try it out and somebody can see it like live here. Then AI, so I am working on these languages Anand, so this will be again, all the information that we Ware capture. It will be coupled into, like, different different language. So you can, like, search based on the language, what are the Coding on in Python or Rust Anand those kind of things. Similarly, what are the hot topics? Speaker 1: Anand, also, I'm trying to put, like, some more, like, to explain about the new users. Okay. What is an MCP? What is a rack? So couple of interesting topics Anand whatever discovery we do based on this GitHub. Speaker 1: So I'm Air to put some AI up about those repositories and how to use it, how it will be helpful for you. And AI, a retain Store insights Anand some kind of use case. So this is currently in progress. So fully, like, Pham trying to wrap up in couple of months. And this is an open source inject. Speaker 1: So anybody who wants to contribute, we have put like all the the details properly Here's how you can contribute to this project. And 1 last thing I would 2026, I would want to say that I am like a supporter of AI 1st. So through this application, I don't track any of the user data. You don't need to sign Anand you don't have to, like, you know, give me your email ID or your phone number or who you are. You're Jonah just go use it, provide your feedback. Speaker 1: Later on, I might put some, like, newsletter kind of things. That's what I'm thinking. But, couple of other projects that I'm currently work. So that is my motto that I don't want to, like, like, any of the user data. I'm not introduces, you know. Speaker 1: You just Code, use this platform, and AI Doug feedback. Thank you so much. Use. Use, I Pham that is what I was sharing. Right? Speaker 1: So just like those agents they they bring that and then, I have 1 weightage. So this is my, signal score Foreman. Right? So it is like today's star plus the number of folks then your 7 day rolling average Anand, what is the activity how active that Repos is. So, you can go through is is like pretty retain I have pretty much put all the details except the code. Speaker 1: So some most of the patterns open source. There are use things AI which is, I haven't open-source yet. So some of those Code algorithms Anand all. But this is overall Alisa this is the weightage. That's how I use it. Speaker 0: That was not what we we need to choose those weights. Oh, this is Speaker 1: so I was so again, as I lead, that there Ware, like, 500 repository. So I was trying little bit here and there, but I think this is and I show you use couple of examples of which which was able to help Week, capture some of the the trending repositories. If you go to all time high, use see like like CLI code AI was able to capture vertices, mem place. So, these all these things I was able to capture my application. So, so far it is working AI, but I am still doing some more explains. Speaker 1: It is very hard to find Find balance. I mean, pretty much you have to, you know, do some trade offs here and there. You cannot like City cannot be like perfect secure that is how the users are, that is how they behave. So, that is how it is. Use. Speaker 1: Any other questions? Speaker 0: Yes. What did you learn about, getting those on GitHub? Speaker 1: Oh, sorry? Speaker 0: What did you learn about how to get those on GitHub? Speaker 1: Oh. Like, what's Speaker 0: their secret? How do they do it? Speaker 1: So GitHub secret you're hosting? Or Speaker 0: Yeah. Speaker 1: Yeah, yeah, that Katz what you need to go there and and click the AI memory. So use will come to know. So then here you can see and then June I am trying to like further add some more Ware more details Here's, but still it will give you a good idea what this is doing and if it sounds interesting to you, you can go ahead and Checkout into the actual GitHub. OSS, do not have AI waste time there, Air know. Speaker 0: I was just wondering if you resolved. Speaker 1: Oh, I definitely Speaker 0: AI. Speaker 1: Use. I I did actually I find some couple of interesting inject. Memplace was 1 of them Alisa like use Node, AI Claude there was a Fire Rate and few other like browsers. I do like lot of scraping Air I said AI background is from data Pham currently Date doing a lot of data engineering work. So I do a lot of scraping projects as well, LLM of analytics projects. Speaker 1: So I I find very interesting projects here, which I use. And fully, AI, I'll, you know, get a chance to do Demos for them as well. Speaker 0: Thank you so much.