AlchemizeCV

Challenge: Transform code repositories into compelling resume content while maintaining technical accuracy—generic AI summaries lack structural understanding. Approach: Built multi-service architecture—Go domain layer handles repository analysis using Tree-sitter AST parsers to extract architecture, API surfaces, and domain features without full compilation. Custom analyzer clusters code by purpose (auth, data, API) and identifies entry points. Python FastAPI backend manages async generation jobs with real-time SSE progress streaming. LLM integration uses structured prompts with diff refinement—initial generation followed by iterative improvements via targeted feedback loops. Key Innovation: Combining static AST analysis with LLM generation produces accurate, quantified bullets grounded in actual code structure. SvelteKit frontend provides Kanban-style job boards, CRUD for experiences/projects, and multi-format export (Markdown, JSON, HTML). Security: Automatic secret redaction scans repos for credentials. Outcomes: Functional CLI with Bubbletea TUI for repo browsing, web dashboard, PostgreSQL persistence via Alembic migrations, comprehensive test coverage.