The problem Future Layoffs solves

Future Layoffs is a revolutionary platform that empowers open-source contributors to discover and tackle bountied issues effortlessly.
A centralized platform for open-source tasks with bounty system, recognition mechanism, and collaboration tools.
Admins can create bounties with a simple comment, eliminating the need for manual setup via a GitHub bot.
A chat feature where users can chat with the repositories.
Established feedback loop for continuous improvement and community engagement programs.

Challenges we ran into

This project has 3 main components . The NextJs full stack application, Github Bot build on Probot framework and the FastAPI server for integrating the LLM model. We mainly faced challenge in navigating the GitHub RestAPI to find what webhooks we were looking for . Also integrating the LLM turned out to be a fair challenge. The FastAPI server would process the whole github repository and then create vector embbedings .

These were the steps we followed to create the FastAPI server :

Index the Codebase: Start by duplicating the repository of interest, loading all files it contains. Then, either split these files for individual indexing or use an existing pre-indexed dataset to skip this step.
Embedding and the Code Store: Utilize a specialized code-aware embedding model to process code segments, which are then stored in the Deep Lake VectorStore for later retrieval.
Construct the Retriever: Implement a Conversational Retriever Chain that queries the VectorStore to identify the most relevant code segments for a given query.
Build the Conversational Chain: Adjust the retriever's settings as needed and apply any specific filters you require.
Ask Questions: Generate a series of questions about the codebase. Utilize the Conversational Retrieval Chain to obtain responses that are sensitive to the context of the query. At this point, a Large Language Model (like GPT-4) will provide detailed answers that take into account both the retrieved code segments and the history of the conversation.