MEET

Inspiration

There are so many ambitious builders, innovators, and creators—especially in this age of AI. However, so many people are restricted to what they can achieve due to their lack of knowledge on events they want to attend.

Sure, sites like luma and eventbrite exist, but they are limited to their own events—not all the events that could possibly be available to them.

In fact, I had to join OpenAI's build week 6 days late just because I didn't hear about it until I overheard it. This is a problem, because there's so much talent in this world that can be uncovered just by providing the right access.

So, i built MEET.

What it does

MEET uses the multiple sources to its avail (web crawling, RSS feeds, eventbrite, list can be expandable) along with the user's general location and preferences to find, personalize, and rank events for the user.

The user will upload their resume, and what they want in events, and the feed will refresh with the latest events the user can attend, quality-checked by AI. This is a real tool that so many people in the world could use to make a real impact.

How we built it

Codex helped me build MEET from an idea into a full-stack app by writing code, fixing bugs, creating the database, running tests, and checking that everything worked correctly. GPT-5.6 helped plan the app, make design decisions, and improve features based on feedback during development. The app uses Groq's Llama model for AI tasks like understanding resumes and matching events, while regular code handles ranking, filtering, and keeping the results accurate and easy to understand.

Built with Next.js, Supabase Auth/Postgres, Exa web discovery, Leaflet maps, and Groq’s low-cost Llama model for resume parsing and semantic event matching. We use deterministic code for distance filtering, ranking, deduplication, and transparent explanations, with RLS-protected user data and a compliant, evidence-backed event pipeline.

Challenges we ran into

• Web crawling and finding events from non-standardized web pages
• Quality-checking events while pulling from the appropriate feeds
• Large delay in pulling events from the internet

Accomplishments that we're proud of

• Pulls from the RSS feeds and Eventbrite, as well as web crawling for more events
• Clean UI with tracing for each event
• Personalization for the events

What we learned

I learned that web crawling is not as easy as just looking online; it's a systematic process that involves much digging and researching. I also learned a lot about managing different API's, making sure no one api is used too much, and how using LLM's for standardized use is a tricky task that involves a lot of prompt engineering and trials.

What's next for MEET

I would want to to expand this into a social network purely for professional events where you can meet people, similar to LinkedIn but for professional events specifically. I also want to add different niches, not just the professional CS / STEM space.

Built With

• codex
• exa
• groq
• next
• supabase