Inspiration
Every quarter, undergraduate students spend hours endlessly clicking through clunky, outdated university directories trying to find a professor whose lab aligns with their interests. When they finally find a match, they stare at a blank screen, terrified of writing the "perfect" cold email. We realized that getting your foot in the door for academic research shouldn't require a masterclass in networking. I built Faculty Connect to solve the "cold start" problem of academic networking by turning intimidating cold outreach into highly researched, warm introductions.
What it does
Faculty Connect is an end-to-end, AI-powered networking agent. A student inputs their major and research interests, and our app scans the faculty directory to find the top three closest matches. Our RAG pipeline vacuums up the professor's personal lab website, synthesizing thousands of words of technical research into a bite-sized "Strategy Dossier." It drafts a concise, sub-150-word cold email which helps the student prepare their introduction,
How we built it
I built the application using a lightweight Python backend and Streamlit for a responsive, modern UI. I used Gemini 3.0-flash-lite as the AI model to assist.
Challenges we ran into
My biggest issue was my naivety upon modern web scraping. When our initial scrapers failed to grab profile links, I had to engineer a creative workaround. We realized UCR's profile URL slugs perfectly matched the first half of a professor's email address. By chopping the @ucr.edu off our static data cache, we algorithmically generated 100% accurate profile links without triggering the university's firewall. What was also funny was the fact that my IP address got soft banned from the BCOE faculty website and I wasn't able to access certain parts of that.
Accomplishments that we're proud of
I'm very proud of figuring out a method around the web scraping issue and this helped me learn a lot about that topic in general.
What we learned
I learned a lot about webscraping and how to connect the Gemini API key to work alongside a project to prompt out and give helpful information.
What's next for AI Research Assistant
Right now, the app is hardcoded for UCR's Bourns College of Engineering. Our immediate next step is to decentralize the data ingestion so any university student can upload their own school's faculty .csv file. I hope to be able to also work with the BCOE website to create an integrated assistant to help direct people around the site.
Log in or sign up for Devpost to join the conversation.