DevContext.AI is an intelligence platform that analyzes GitHub repositories to generate employability scores, code-quality metrics, and mock interviews. The system uses a grounded AI approach, referencing specific files and line numbers to avoid hallucination, and distinguishes between developer code and boilerplate framework code. It utilizes a multi-model strategy via Amazon Bedrock (Claude Haiku for fast reviews, Sonnet for deep architecture analysis).
3-stage pipeline: Project Review (~30s), Intelligence Report, and Interview Simulation.
AI claims are grounded in actual repository evidence (file paths and line numbers) to prevent hallucination.
Designed for 10+ concurrent analyses, 50MB repositories, and >50K tokens.
React frontend, AWS SAM serverless backend, DynamoDB, and WebSocket protocols.