I'm an AI & MLOps Engineer focused on LLM Security — I build LLM systems and the secure, observable infrastructure that keeps them in production: deployable, private, and hard to break.
At Youlex (legaltech), I contributed to a hybrid-RAG legal search engine (BM25 + dense vector + RRF + BGE reranker over TCU case law with Legal-BERTimbau), reaching Recall@5 of 94% — plus the MLOps around it: an 8-stage pipeline with zero-PII sanitization and a PII-safe staging environment with an automated safety gate. As an independent consultant, I shipped WhatsApp AI agents to production (n8n, VPS, Docker, Postgres/Supabase, Redis) with human handoff and observability, cutting response time from 6–12h to 1–3min at 24/7 uptime. At Outlier, I work freelance on LLM evaluation — designing agentic eval tasks and scoring model trajectories, including prompt-injection testing.
I also lead Ares, a cybersecurity / DevSecOps extension project at UFC. Currently preparing for the CKA (Certified Kubernetes Administrator).
Feel free to explore my repositories and follow my progress.



