← all projects

Kairo

Zero to shipped in a week

An AI patient assistant that talks back, in text and in voice. I took it from an empty folder to a shipped product in under a week, working entirely AI native.

  • Python
  • FastAPI
  • React
  • TypeScript
  • OpenAI
  • Vapi
  • AWS EC2
github ↗

Problem

Patient-facing healthcare tools are usually slow to build and clunky to use. The goal: a working LLM-powered patient service experience — chat plus voice — shipped fast enough to validate the idea.

Approach

  • FastAPI backend orchestrating OpenAI for the conversational layer, with Vapi handling voice-call handoff.
  • React + Vite + TypeScript frontend.
  • Deployed on AWS EC2.
  • Built AI-native: Cursor and Claude Code drove the development loop, which is how a one-person team ships a full-stack LLM product in days.

Result

End-to-end patient service app — chat, voice handoff, deployed — 0 → shipped in under a week. A case study in how much one engineer can ship with an AI-native workflow.