GitResume
An AI-assisted tool that reads a repository and turns real implementation evidence into clearer, job-specific resume language.
GitResume came from a problem I kept noticing: strong engineering work often gets described weakly.
The repository shows real decisions, tradeoffs, and implementation detail. The resume bullet often says something generic.
GitResume tries to close that gap without inventing achievements.
What It Does
The tool takes two inputs:
- a GitHub repository
- a target job description
It inspects the codebase with Tree-sitter, infers structure and stack, then generates resume bullets, project summaries, and interview questions grounded in the repository.
The job description acts as the lens. The codebase stays the anchor.
What Mattered
The hard part was not making an LLM write. The hard part was extracting enough signal from real code that the output felt reviewable.
That led to a few product decisions: Tree-sitter instead of shallow keyword matching, WebSocket streaming for long analysis, GitHub OAuth for private repos, Redis for transient session state, and modular model support.
Why It Stayed
GitResume is strongest when it stays opinionated.
It should not become generic career software. It should behave like a technical writing assistant for engineers: evidence first, wording second, always tied back to what the repository actually proves.