Pietro Ferragamo

Applied AI Engineer. I build and evaluate LLM systems in production. Writing about evaluation, interpretability, and things I find interesting.

Side A Selected work all projects

Behavioral evaluation framework for agentic LLM systems

CSI Piemonte

Two-stage search engine for public tenders

CSI Piemonte

Internal MCP server for agent tool composition

CSI Piemonte

AI Voice Assistant

Personal project
Side B Writing all posts

GenAI assistants in the public sector: measuring safety and trust

Measuring safety and trust in a national-scale PA assistant, from daily red-teaming scenarios to per-tender behavioral evaluation.

Agenda DigitaleJuly 10, 2026 · 3 min

AI assistants in the public sector: extending them with reusable tools

Extending production AI assistants with reusable primitives and declarative composition, instead of rewriting a tool every time.

Agenda DigitaleJune 22, 2026 · 2 min

Camilla, the AI assistant taking on Italy's bureaucracy

Architecture and governance choices behind a national-scale chatbot for Italian Public Administration.

Agenda DigitaleApril 30, 2026 · 2 min