Exam Grader

Inspiration

bad grading by profs, pretty sure AI can do it better despite its ton of flaws.

what i learned

spent the whole day reading through the langchain documentation, learned how to implement RAG properly and what sort of decisions to take when making a RAG application. Will try out langgraph next, was seeing langsmith as well but well... paid service and didnt seem useful for now.

how i built

as i said, read through the langchain documentation the whole day, looked for what seemed right and tried to integrate that into my code properly, got tripped up with some google cloud SDK thing in between but realised it wasn't needed and moved forward.

challenges i faced

tbf not much since it was a simple RAG application and most of the code was right there in the documentation. however, i dont know how to build a UI so.... will try to learn a little of webdev before the next hackathon so that i can do that by myself too. also using AI to code the UI messed up my whole model as well somehow and i had to fix it somehow by myself so defo not letting that happen again.

Built With

• chromadb
• google-gemini
• langchain
• pymupdf
• python
• streamlit