AI · SaaS
Podding
AI-driven podcast guest booking, end to end.
- Role
- Senior Full-Stack AI Developer (solo build)
- Period
- 2024 – 2025
- Links
- Client site ↗
Outcomes
Reduced end-to-end workflow time by 85–88%
95% podcast acceptance rate via semantic matching
Fully automated outreach and response tracking
The problem
Booking podcast guests is a manual grind: researching shows, matching topics, outreach, and follow-ups consume hours per placement and most pitches go unanswered.
The approach
Built the platform solo: LangGraph-orchestrated agents over a RAG pipeline with vector-database semantic matching between guests and shows, Rephonic API for podcast data, automated email outreach with response tracking, and real-time updates over Socket.IO. Django/Celery backend, Next.js + Shadcn UI frontend, deployed across Vercel and AWS Lambda.
Stack
- Next.js
- Shadcn UI
- Django
- Celery
- Socket.IO
- LangGraph
- PostgreSQL
- RAG
- Vector DBs
- Vercel
- AWS Lambda
Under NDA. The product interface is confidential. Shown here are public marketing materials and the platform login only. Outcomes above are publicly listed figures.
