Inspiration

Automate lending & compliance.

What it does

ML risk scoring (DTI logic) & RBI RAG.

How we built it

Streamlit, FAISS, Databricks Serverless.

Challenges we ran into

MLflow schemas & LLM math hallucinations.

Accomplishments that we're proud of

Streaming 70B LLMs with low latency.

What we learned

Decoupling UI & compute via REST.

What's next for Arth-Sarthi

Continuous volume ingestion & multilingual.

Built With

Share this project:

Updates