Developer Builds Local AI-Powered Mock Interviewer for Rust Code

Software developer Yash K. Saini has created 'grill', an open-source, local-first mock technical interviewer designed to help developers practice for coding interviews using their own repositories. Built for a friend, the tool uses Google's Gemma model via Ollama and the Mastra agent framework to analyze codebases for complex patterns like unsafe blocks, locks, and async operations. Unlike traditional question banks, grill scans a user's actual projects to generate context-aware interview questions, follow-ups, and performance grades. By running entirely on the local machine, the tool ensures that sensitive or private code remains secure while providing a personalized practice environment. The project emphasizes the benefits of open-weight models and local execution, allowing users to customize rubrics and iterate on their interview skills without relying on external cloud APIs or worrying about data privacy.