I am a Senior Machine Learning Engineer at Thomson Reuters based in Zug, Switzerland 🇨🇭.
As the inaugural ML expert within the Cybersecurity team, I drive the strategy, development, and deployment of AI-based tools to enhance the company's cybersecurity posture and secure AI products.
My core focus and responsibilities include:
- Vulnerability Remediation: Designing and building tools to discover code vulnerabilities and plan automated remediations.
- LLM Security: Leading research and engineering efforts dedicated to securing LLM-based applications.
- AI Subject Matter Expert: Serving as an internal consultant and SME on AI-related projects across various cybersecurity teams.
- Agentic Frameworks: Building frameworks to scale the development of agentic projects (focusing on AgentOps, evaluation harnesses, and loop engineering).
Previously, I was an Applied AI Researcher at Thomson Reuters Labs working on LLM fine-tuning and machine learning systems for Tax, Legal, and Global Trade products.
My journey includes:
- Pi School (School of AI): Awarded a scholarship (1 of 10 fellows) to research prompt engineering and parameter-efficient fine-tuning (PEFT) on LLMs (GPT, BLOOM, Flan-T5) for Adaptive Machine Translation in partnership with Translated.
- SafEcho AG: Developed LLM conversational assistants for Occupational Health and Safety compliance.
- Sapienza Università di Roma: Graduated with an MSc in Engineering in Computer Science.
I collect my structured research notes, paper summaries, and guides covering Machine Learning, MLOps, AI Safety, and Cybersecurity (and other non-tech topics) in my [Obsidian Vault(https://gianfree-notes.netlify.app/).
Get in touch:
- Email: gianfranco.romani@outlook.it
- 📄 Check/download my resume here
Beyond the screen:
- Swimming
- Building mechanical keyboards
- Beating escape rooms
- Tasting specialty coffee
- Cooking