SynchroSched

landing page

About the Project

Inspiration
Coordinating study sessions with friends was always messy — too many group chats, doodle polls, and last-minute changes. We wanted something where you could just say out loud “find us 2 hours tomorrow after 3pm” and instantly get a booked time that works for everyone.

What We Learned
We learned how to design a full-stack scheduling system on a tight timeline, handle natural language parsing for time constraints, and integrate with AWS DynamoDB for reliable, serverless persistence. We also gained experience connecting multiple APIs (voice, LLM, Google Calendar) into a single streamlined workflow.

How We Built It

Frontend: Next.js 14 (App Router) + TypeScript + Tailwind for a fast, clean UI.
Backend: API routes in Next.js backed by AWS DynamoDB, storing groups, members, and calendar events.
Scheduling: Custom overlap algorithm that merges busy intervals across members, inverts to free time, and suggests the top available slots.
Parsing: Regex fallback to convert transcripts into constraints, with hooks ready to swap in Groq LLM for richer understanding.
Integrations: Designed for VAPI (voice → text) and Google Calendar (push events directly to members’ calendars).

Challenges

Modeling flexible calendar data in DynamoDB and ensuring queries stayed fast.
Getting natural language like “tomorrow evening” to map cleanly into hours and dates.
Handling time zones and making results feel intuitive across different members.
Balancing hackathon speed with production-ready extensibility.