TechOnboard is an AI-powered technical onboarding platform designed for engineering teams. When a manager adds a new developer, they select an AI agent (specialized by role: backend, frontend, DevOps, QA, etc.) and the developer's seniority level. The platform then conducts a real-time streaming interview via WebSocket powered by Google Gemini, analyzing the developer's GitHub profile to detect their actual tech stack, identify skill gaps, and understand their learning style. Once the interview is complete, a LangGraph pipeline runs in the background through five stages: GitHub profile analysis, access provisioning (GitHub, Jira, Slack) enforced by a Lobster Trap security policy engine, RAG-based codebase indexing and tour generation, intelligent Jira ticket assignment matched to the developer's level, and finally a Gemini-generated personalized onboarding plan. The manager dashboard provides full visibility into all active onboarding sessions with real-time status tracking, audit logs of every security policy decision, and the ability to cancel or delete sessions. Payments for agent activation are handled via the x402 protocol on Base Sepolia testnet (USDC). The frontend is built with React, TypeScript, Vite, and TailwindCSS, served via nginx on Railway. The backend runs FastAPI with async SQLAlchemy on PostgreSQL with pgvector for RAG embeddings, Redis for caching, and is fully deployed on Railway with four services: frontend, backend, PostgreSQL, and Redis.
Category tags: