Mocki

Multi-agent AI mock interview platform powered by NVIDIA Nemotron.

Mocki runs a coordinated pipeline of specialized agents — a Coordinator that plans the interview flow, a Persona generator, an Interviewer, a Clarifier that detects vague answers and asks follow-ups, an Evaluator that scores each response, and a Reporter that produces the final assessment. A live agent activity panel surfaces every agent's reasoning, decisions, and token usage in real time.

Stack

• TanStack Start (React 19, file-based routing, server functions)
• Vite 7 + Tailwind CSS v4
• NVIDIA Nemotron (nvidia-nemotron-nano-9b-v2) via the NIM API
• Cloudflare Workers runtime for production SSR

Getting started

bun install
echo "NVIDIA_API_KEY=your_key_here" > .dev.vars
bun dev

Open http://localhost:8080.

Scripts

• bun dev — start the dev server
• bun run build — production build
• bun run preview — preview the production build
• bun run lint — lint the project

Project layout

src/
  routes/           # File-based routes (TanStack Router)
  server/           # Server functions and agent pipeline
    agents.server.ts        # Coordinator, Interviewer, Clarifier, Evaluator, Reporter
    interview.functions.ts  # Orchestration entrypoints
    nim.server.ts           # Nemotron client
    agent-log.server.ts     # Per-session activity log
  components/       # UI components
  lib/              # Client store and utilities

License

MIT