Leone Lage Perdigão

Path

My path started in software engineering in Brazil in 2010. Over fifteen years I moved through full-stack development, cloud architecture, and platform engineering, leading teams at IBM, PostNL, Tikkie (ABN AMRO), and Randstad. The shift into AI came from a specific frustration: the gap between what AI can do in research and what it actually delivers in production was where the most interesting projects were getting stuck. I went back to university for an MSc in Artificial Intelligence (Distinction) from the University of Bath, UK, while continuing to lead AI engineering teams in industry.

TaoQ AI is the independent practice I founded — an AI practice spanning the full value chain, with sharper focus on security, governance, and evaluation. The clients are European technology companies; the work is concrete — threat modelling, agent red-teaming, conformity gap analysis, Annex IV technical documentation, evaluation frameworks. I work with engineering leaders on the questions that don't fit neatly on a slide: where their security posture is actually exposed, when to build and when to buy, what compliance looks like when you have to defend it in front of a regulator or an auditor. The engagements stay close enough to the work that I remain in the code with them.

I was previously Lead AI Architect at PostNL's AI Centre of Excellence. I authored the PostNL AI Reference Architecture — adopted as the organisational standard for taking generative and agentic AI from prototype into evaluated, governed, live deployment — and led three teams across the AI value chain (Prototyping, Core Tech & R&D, and AI Adoption & Governance), working alongside the CTO and Principal Engineers on the technology choices that mattered at the scale of a 30,000-person organisation. The MLOps/LLMOps standards were built to meet EU AI Act expectations; the goal is the one I still hold — AI that is not only capable, but defensible.

Before that, as Advisory AI Engineer at IBM, I was the technical authority for AI solutions across Northern, Central, and Eastern Europe — architecting enterprise systems on WatsonX, fine-tuning foundation models for strategic clients, and leading the AI side of IBM QRadar EDR. Earlier still, as Lead Software Engineer in PostNL's Cloud Centre of Excellence, I architected the serverless-first AWS Cloud Platform that now serves 100+ teams across 1,000+ AWS accounts. Further back: cloud-native API engineering at Tikkie (ABN AMRO), the serverless AWS platform at Randstad Holding that integrates its 40+ operating companies globally, and Solutions Architect at Banco Inter, Brazil's second-largest digital bank.

Mentorship is part of how I stay sharp: I'm an IBM Blue Core Mentor and an active MentorCruise tech mentor (5.0 rating). The academic foundation behind the practice is the MSc from the University of Bath (2022–2025 — Machine Learning, Reinforcement Learning, Robotics & Machine Vision), after Computer Engineering at CEFET-MG and a BSc Information Systems at Universidade FUMEC, Brazil.

Open Source & Research

I created Ziran, an open-source AI agent security testing framework (Apache-2.0) covering tool chain vulnerability analysis, multi-phase exploit campaigns, and CI/CD security gating. I also contribute to NVIDIA's Garak (LLM vulnerability scanner) and Hugging Face's Transformers, and I am building open-source tooling for post-quantum cryptography — quantum threat modelling, cryptographic agility tracking, and CI/CD compliance scanning for quantum-vulnerable algorithms. The hands-on work keeps the architectural decisions honest.