A Brief Note

I was looking through hackathon websites to hone my skills and learn something new when I came across this hackathon on DevPost. It was relatively small and utilized AI and seemed to be an interesting challenge as it was using Modus and Hypermode, two relatively new frameworks.

Modus was fast, and initially I had no clue on what could be done with it, so I just decided to set it up and play around with it. When I read through the rules, two things struck me - We had to use Knowledge Graphs and either Go or AssemblyScript.

I thought it would be a great way to learn Go and decided to go with that and started with Neo4j as the Graph Database. Initially the plan was to create an app that would let Recruiters search through and filter and automatically cold call proficient candidates but then I realised it would not be done by the time I finished the hackathon, so I had to pivot my idea altogether.

This time I saw a recently published open-source Github Repository on GitIngest, where the maker made it easy to provide context to LLM applications for Github Repositories. This, combined with the fact that I really hated clicking on the tiny button for followers and following in Github to find Repositories, made it so that I created GitExplore using Github API's Dgraph and Hypermode LLM's.

GitExplore would essentially allow users to view their friends and followers and their repositories and starred repositories and chat with any one of them and get context driven answers which would be extremely useful and fast.

Conclusion

GitExplore won the best use of Dgraph Award and it motivated me to work on it as a nice side-project to help myself and other students just getting into Github and exploring.