Inspiration
Our inspiration for Maargdarshak came from the capabilities of AI/ML, similar to ChatGPT, to transform complex information into easily accessible knowledge. We aimed to create a solution that simplifies navigating through the ICDS documentation.
What it does
Maargdarshak answers questions prompted by users that are relevant to Penn States ICDS (Institute of Computational Data Science) user guide.
How we built it
We mainly used Rust and Python for the full stack, streamlit framework for the frontend, OpenAI API, and JSON to store the chat history.
Challenges we ran into
For the frontend we struggled to convert HTML and CSS to the streamlit framework. For the backend, we had trouble figuring out the correct model to use for the ideal output. Additionally, we had to search and modify rust dependencies for generative AI and figure out how to web scrape the correct information through noticeable patterns.
Accomplishments that we're proud of
We were proud that we successfully implemented an LLM and that we efficiently produced accurate responses using open AI's API.
What we learned
We learned how to implement streamlit for the frontend and how to interact with models using Rust.
What's next for Maargdarshak?
We plan on integrating this bot for any Penn State webpage. By changing the source text, we can implement this bot for any possible text form of information.
Log in or sign up for Devpost to join the conversation.