TypeScript
Next.js
React
Tailwind
MongoDB
Pinecone
LangChain

Overview

Challenges & solutions

Challenge

Accurately parsing and extracting content from uploaded files involved various file formats with unique parsing intricacies. Handling temporary file uploads and precise deletion post-conversion in Vercel deployment was crucial.

Solution

I created API routes for uploading files to a temporary directory and extracting document content using open-source or LangChain document loaders. File content was converted to vector embeddings, stored in Pinecone and MongoDB namespaces, and temporary files were deleted.

Project Chatbot
Project Chatbot
Project Chatbot
Project Chatbot