Inspiration
Its often random if you might the right person at a hackathon or a business event. We want to use the power of AI to make sure we meet, connect with and build with the people most aligned in terms of interest and compatible in terms of roles.
What it does
AI agent recommends the best team members Upon arrival, participants respond to a few questions LLM creates a structured profile for each participant and stores it in the database Agent recommends the best matches based on the structured data Agent summarizes and scores matches for each participant based on match strength
How we built it
We created a front end app in react with questions and an API. We sent the profiles to the database and used Mongadb vector to do a semantic search to score people based on similarities. Using LLMs, we are extracting relevant information and sending the profiles back to the front end. The top matches with scores based on how well they matched are shown to each participant.
Challenges we ran into
Refining our matching algorithm takes a lot of data and experimentation. Not enough time to integrate different solutions together for a great demo.
Accomplishments that we're proud of
We can match people based on their project ideas, their interests, their roles, functions and backgrounds!
What we learned
We learned that Mongadb is very effective in vector search
What's next for KinConnect
-Import external data sources about participants LinkedIn Github -Scale to use at conferences with 30,000+ people.
Log in or sign up for Devpost to join the conversation.