architecture
survey

📚 Project Story – “Event Assistant Agent”

✨ Inspiration

San Francisco pumps out 50-plus tech gatherings every week. Picking the one that’s worth burning an evening on feels like doom-scrolling LinkedIn meets FOMO roulette. We imagined a sidekick that scores every event, so you spend 5 seconds deciding, not 50 tabs comparing.

Confirmation, we run a survey to confirm the need.

video: https://drive.google.com/file/d/1N1TcF_X6L37JN5p9PQhnNp3cSRllNzk9/view?usp=sharing

survey: https://events-survey.lovable.app/

github: https://github.com/ottomansky/event-matcher

Vibe description: Project Overview (as mapped in the diagram) Goal: Rank tech events (meet-ups, hackathons, workshops) by a composite Event Quality Score so professionals in San Francisco instantly see which gathering is worth their time.

Data Sources Source Role Lu.ma Events Primary feed of tech meet-ups & conferences. Eventbrite Public events & workshops. Meetup.com Community gatherings. DevPost / MLH Hackathons & competitions.

Each source streams raw Events into the collection layer.

n8n Collection & Processing Component Purpose Event Scraper Hub Parallel scrapers (one per source) with URL deduplication. Enrichment Queue Priority-based jobs that call external APIs for extra facts. Scoring Engine Applies multi-factor analysis to produce the Event Quality Score.

Processing is orchestrated in n8n; jobs flow Scraper → Queue → Scoring.

Enrichment APIs API Data pulled Factor it feeds Crunchbase Company valuations & funding rounds Sponsor credibility GitHub API Repo stars, contributions Speaker / project clout LinkedIn Follower counts, job titles Networking crowd Google Places Venue ratings & geo Venue vibe Glassdoor Employer reviews Sponsor reputation
Storage & Cache Store Responsibility Supabase Source-of-truth relational DB for events & scores. Redis Cache Low-latency store for recent API responses and hot scores. Vector DB Embedding index of events (for semantic search & alert matching).

The Scoring Engine writes to Supabase and pushes embeddings to the Vector DB; Redis accelerates repeated look-ups.

Output Interfaces Interface Audience / Usage Webhook API Real-time scoring requests from partner apps. MCP Server AI-tool endpoint so agent frameworks can “call” the event scorer. Web App Interactive map / list used by end-users. Notifications Email / SMS alerts when a high-score event matches user interests.

Alerts originate from the Scoring Engine → Vector DB similarity search → Notification service.