AI · Marketplace
Kayana
A hiring marketplace that runs itself.
- Role
- Full-Stack AI Developer
- Period
- 2022 – 2023
Outcomes
Reduced hiring time by 90%
Improved matching speed by 85% via semantic skill matching
Recruitment flow runs with minimal manual involvement
The problem
Matching virtual assistants with employers required manual screening, skill assessment, and pairing: slow, inconsistent, and impossible to scale.
The approach
Built the marketplace SaaS from scratch and designed a LangGraph multi-agent system that automates the recruitment flow: semantic skill matching over Pinecone embeddings, automated screening, and Stripe-powered payments on a Django + Next.js stack.
Stack
- Next.js
- Django
- LangGraph
- AI Agents
- Pinecone
- Celery
- Stripe
- PostgreSQL
- AWS
