Problem Description

Create a typescript langchain app that takes 10 transcripts from a folder, creates embeddings for them with huggingface + langchain, stores them into a supabase postgres vectorDB and then queries this data to return a summary of summaries with GPT. I can provide you with the 10 transcripts.

Acceptance Criteria

Input: The app should be able to read 10 transcripts from a specified folder.

Embeddings Creation: The app should use HuggingFace and Langchain to create embeddings.

Storage: The app should store the generated embeddings into a Supabase PostgreSQL VectorDB.

Querying Data: The app should be able to query the stored embeddings data from the VectorDB.

Summary Generation: The app should use GPT (Generative Pre-trained Transformer) to generate a summary of summaries based on the queried embeddings data.

Output: The app should return the generated summary of summaries as output.

TypeScript: The entire application should be written in TypeScript.

Documentation: The app should include clear documentation on how to set up, configure, and run the application.

Testing: The app should include unit tests to ensure proper functionality and validate the acceptance criteria.

Error Handling: The app should handle errors gracefully and provide informative error messages when necessary.

Technical Details