EVDocent

• 

  EV Docent Logo

Inspiration

Many people would like to rent EVs but are unsure of which models are most suitable for them

What it does

EVDocent recommends EV models based on the user's requirements, thus ensuring the user does not have to compromise because of choosing the wrong EV models

How we built it

We created an EV review crawler which stores data into Snowflake. We then used Mistral-large LLM to perform RAG on the EV reviews based on the user's inputs

Challenges we ran into

We had initially created the app in the Snowflake environment and faced a challenge when migrating it to the Streamlit cloud

Accomplishments that we're proud of

We are very proud of the contextuality of the answers provided by the chatbot (based on our tests so far). The chatbot also does not hallucinate and is able to clearly state the limit of its knowledge

What we learned

Snowflake is a very handy platform to create LLM-based applications and has a very easy-to-use user interface

What's next for EVDocent

We would like to integrate with external APIs (such as charging station operators) to better help EV renters understand where to charge their EVs during their trip

Built With

• mistral
• python
• snowflake
• streamlit