Hitachi Challenge - Ball Knowers

Inspiration

Was introduced to the problem from Datathon

What it does

Detects PII and sensitive information from PDFs

How we built it

Used Hugging Face Models and Python. Python was used to send requests to a Hugging Face LLM model (Kimi-K2-Thinking), as well as to process PDFs to extract text and convert to images, providing data to feed to the LLM

Challenges we ran into

The model is pretty slow, and we ran out of time to create a web interface and to create a demo

Accomplishments that we're proud of

Creating a prompt for the Kimi-K2-Thinking model and interfacing with Hugging Face's platform to interact with LLM models.

What we learned

Learning how to interface with Hugging Face's platform to interact with LLM models. Was great learning about the open source LLM space!

What's next for Hitachi Project - Ball Knowers

Creating a web UI and finding ways to optimize in order for it to be faster

Built With

• huggingface
• python