Bridge

Inspiration

One of the most common advice older students give is to join more societies. We personally noticed how difficult it was to find and keep track of events designed for students that can contribute meaningfully to their career.

What it does

A single unified source of information to discover student career events, tailored to students for their major, interests and year level. It searches through external websites to find events, including society, extra-university societies, companies, and university program websites to push tailored events and programs as well as their dates.

How we built it

Bridge is built on a modern full-stack architecture using Next.js with tRPC, Supabase as our database, and shadcn/ui for the interface.

On the data side, we used Cheerio to scrape a curated set of university and society websites, pulling live event and application data directly from the source. That raw data is then processed through Gemini, which parses each entry and generates structured tags, categorising by event type, relevance, and opportunity category, before being stored in Supabase.

From there, the personalisation layer kicks in. When a student completes onboarding, telling us their degree, major, year level, and interests, Bridge queries the database and surfaces only the events and opportunities that match their profile.

The dashboard presents this as a clean chronological timeline, showing event details, the source, and a direct link out to the original listing. A separate tab surfaces application opportunities like research programs and society recruitment, ranked by fit to the student's profile.

Challenges we ran into

The biggest challenge was building a reliable data pipeline under time pressure. Scraping different websites means dealing with inconsistent HTML structures, and getting Gemini to produce consistent structured tags required careful prompt engineering. We made a deliberate decision to scrape a curated set of high-quality sources rather than scrape broadly and deal with noise.