AI Engineer. I build AI agents, agent harnesses, RAG pipelines and MCP servers, and the full-stack products around them.
Postgraduate student in Generative AI Engineering (Data Science Academy, 2026–2027), with 10+ years shipping web applications in React, Next.js, TypeScript, Node.js and Python.
- Applied AI: LLMs, AI agents, agent harnesses, RAG, MCP, LangGraph, OpenAI API, embeddings, vector search, LanceDB
- Full stack: Python, Node.js, REST APIs, BFF, Docker, CI/CD
- Frontend: React, Next.js, SolidJS, TypeScript, Tailwind CSS, Vite
- Quality and architecture: Vitest, Playwright, Testing Library, Clean Architecture, DDD
- vault-search-mcp - Python MCP server for hybrid search (vector + full-text on LanceDB, reranking, graph navigation) across Obsidian vaults and Markdown knowledge bases.
- llm-router - local LLM router for OpenCode, with adaptive routing across MiniMax, GLM, Claude and Codex and configurable tool policies.
- generate-commit-extension - VS Code and VSCodium extension that writes Conventional Commits from your staged diff using LLM providers.
- carreira-guiada - agent workflow in Python that automates the job search: ATS résumé, job matching, form filling, applications and e-mail follow-ups, running on Claude Code, Codex CLI or the model you prefer.
- content-lens - local-first browser extension that filters algorithmic feeds with explainable rules, plus opt-in AI classification and AI-assisted rule creation.
- cdxt - interactive on/off toggle for MCP servers and plugins in the Codex CLI config.
