Summary
António De Castro is a Senior Machine Learning Engineer with a PhD in AI and over a decade of experience designing production-grade, safety-critical decision systems for regulated domains like aviation. He specializes in system-level intelligence—architecting agentic AI and multi-agent decision platforms that combine LLMs, reinforcement learning, optimization, and robust RAG/retrieval strategies to deliver reliable decisions at scale. As a founder and staff/principal-level engineer he has led end-to-end delivery from problem formulation to CI/CD-ready deployment, building LLMOps/AgentOps frameworks with Databricks, MLflow, Docker and Kubernetes. His work uniquely blends applied research and teaching with enterprise engineering, turning multi-agent coordination and RL research into operational platforms used in airline disruption management. Based in Vila Nova de Gaia, Portugal, he is particularly focused on decision-making under uncertainty and evaluation beyond static benchmarks, often prioritizing traceability, safety, and operational constraints over raw model metrics. An early practitioner of agent negotiation and constrained RL in production, he brings rare domain depth bridging aviation operations and modern agentic AI.
8 years of coding experience
20 years of employment as a software developer
Postgraduate Astrobiology, Postgraduate Astrobiology at Universidade Nova de Lisboa
Postgraduate Air Transport Operations, Postgraduate Air Transport Operations at ISEC - Instituto Superior de Educação e Ciências
Informatics, Informatics at Leeds Beckett University
High School Diploma mechanotechnics, High School Diploma mechanotechnics at Escola Industrial Infante D. Henrique
5 Year Undergraduate Information Systems Engineering, 5 Year Undergraduate Information Systems Engineering at Instituto Superior de Engenharia do Porto
PhD in Informatics Engineering Artificial Intelligence, PhD in Informatics Engineering Artificial Intelligence at Faculdade de Engenharia da Universidade do Porto
English, French, Portuguese