AI-Powered GitHub Repository Analyzer

Overview: Co-developed a high-performance, full-stack software application in a student team collaboration designed to ingest, parse, and analyze entire GitHub repositories. The platform acts as an automated codebase architect, mapping repository structures and generating contextual engineering summaries using state-of-the-art LLMs.

Technical Architecture & Workflow

• Full-Stack Architecture: Collaborated on architecting the user interface and analytical backend using the Python/Shiny reactive framework for seamless asynchronous operations.
• API & Ingestion Orchestration: Programmed a robust integration with the GitHub API to securely pull, walk, and deserialize directory trees, file structures, and source code files.
• LLM Ingestion Pipeline: Configured and tuned the Gemini 1.5 Flash LLM, engineering optimized system prompts to parse complex multi-file code dependencies without breaking token limits.

Engineering Capabilities Demonstrated

• Codebase Mapping: Translates fragmented, undocumented software trees into highly structured, scannable technical intelligence reports.
• System Automation: Drastically reduces codebase onboarding friction for engineering teams by automating structural architectural reviews.