At IBM, I lead a five-engineer squad delivering the Trenitalia AI ecosystem for critical railway infrastructure. The systems sustain 99.9% availability, and I designed and shipped a multi-agent QA system that automates 95% of regression testing.
For Abivet, I built VetAnalytics end to end, an AI veterinary-diagnostics platform now used by clinics. It turns blood-test PDFs into structured diagnostic reports using GPT-4o Vision.
At Scale AI, I built RLHF pipelines for LLM code generation and designed evaluation frameworks and QA protocols for training datasets. I also engineered event-driven microservices for fine-tuning data and internal dashboards for monitoring training progress and data integrity.
I work across AI systems and production software, from agentic workflows and distributed backends to user-facing applications. My open-source work includes uuid7-typed, a type-safe UUIDv7 generator, and an interactive 3D portfolio built with React and Three.js.

