Diego Lugo

At Finva, I shipped a hybrid credit-risk ML decisioning system using causal analysis, XGBoost, neural networks, and threshold calibration that sustains 85% accuracy in production. I also build Python backend services, optimize APIs and database queries, and translate model outputs into actionable recommendations for credit advisors.

For a tax audit SaaS, I built an API-based LLM-agent workflow that reconciles invoice and payment data and generates audit commentary from fiscal rules and legal context, reducing manual review from roughly eight hours to five minutes per case. I’m moving deeper into AI engineering across RAG, agentic systems, and MLOps.