LumedIn - a multi-agent, 1-stop solution that streamlines the process of finding collaborators for hackathons

Inspiration

Finding the right teammates at hackathons is a critical yet painful process. With hundreds of participants, each with unique skills and interests, manually identifying complementary collaborators becomes overwhelming. We recognized this universal pain point and built LumedIn to transform hackathon networking from chaos into clarity.

What It Does

LumedIn is a multi-agent system that automates hackathon team formation through intelligent matchmaking. The platform:

  • Scrapes Luma event attendee lists and extracts LinkedIn profiles via API
  • Employs GPT-powered agents to analyze participant skills, experience, and interests
  • Generates compatibility rankings based on complementary skillsets
  • Provides AI-drafted outreach emails to facilitate introductions
  • Includes facial recognition to help users identify participants in person

How We Built It

Our architecture consists of five coordinated stages:

  1. Data Collection: Chrome extension scrapes Luma attendee data
  2. Profile Enrichment: Scraper API retrieves LinkedIn information
  3. Intelligent Ranking: OpenAI API analyzes profiles and generates compatibility scores
  4. Dashboard Display: Next.js frontend presents ranked results with detailed justifications
  5. Outreach Automation: AI agent drafts personalized emails

Our bespoke ranking algorithm evaluates skill complementarity, shared interests, and experience levels to optimize team composition.

Challenges

  • API Rate Limits: LinkedIn scraping required careful request throttling and error handling
  • Ranking Algorithm: Balancing multiple factors (skill diversity, experience, interests) into a single coherent score
  • Real-time Performance: Processing hundreds of profiles within acceptable latency bounds
  • Privacy Considerations: Handling personal data responsibly while providing useful insights

Built With

Share this project:

Updates