M.S. Data Science @ EPFL & ETH Zurich | Software Engineer | NeurIPS 2025 Spotlight Benchmark (AI Agent Security)

AI & Machine Learning Research

My research targets the critical areas of AI robustness and security.

• AI Agent Security: I investigated vulnerabilities in Computer Use Agents, with findings published as a spotlight paper of the NeurIPS 2025 Datasets & Benchmarks track, and at the ICML 2025 workshop on Computer Use Agents. My work established a new benchmark for agent security, as well as automated judges to catch and prevent harmful agent actions.
• AI Safety Leadership: As the lead of Safe AI Lausanne, I manage a community of 20+ members. I coordinate research projects, organize hackathons, and spearhead reading groups exploring topics from evolutionary AI systems to diffusion models.
• Large‑Scale AI Systems: I contributed to the Retrieval‑Augmented Generation (RAG) pipeline for the Swiss AI Initiative. I started a student research group who contributed to Switzerland’s sovereign and fully open LLMs, helping to build a scalable, nation‑wide language model infrastructure.

Software Engineering & Open‑Source

I have practical experience building and maintaining software at scale.

• Startup Experience (Acquired by Airbnb): As a Software Engineer Intern at Urbandoor, I shipped production front‑end features using React and TypeScript, contributing to the growth of the platform prior to its successful acquisition by Airbnb.
• Full‑stack Engineering: While interning at Adaptyv Bio (biotech startup), I developed full‑stack production apps with Next.js, and engineered data processing pipelines in Python. I addressed the customer’s most common requests by building a detailed product catalog, and I saved the co-founders dozens of hours by automating internal processes.
• Open‑Source Maintainer: As a maintainer for the FreeCodeCamp Guide, I was responsible for integrating new content into a learning resource used by millions of developers worldwide. I contributed to many other projects, from the OSWorld environment for AI agents to EPFL’s course review website used by hundreds of students.

Teaching & Leadership

I'm passionate about making complex topics accessible and empowering others. I have:

• Developed and taught a git & GitHub workshop to over 250 first‑year CS students at EPFL.
• Helped over 300 students at EPFL learn and organize more efficiently through in‑person workshops, and hundreds more through my YouTube channel and online course.
• Individually coached a dozen of students to help them reach their goals, build better habits and study systems.
• Hosted podcasts and radio shows at EPFL’s student radio, did team‑building at the sustainability club, and worked for TEDxGeneva.

How I work

• Bias to deliver: I ship in small, reliable increments, backed by solid tests.
• Impact‑driven: I move fast, focus on leverage, and aim for measurable results.
• Autonomous: I dive into problems independently, prepare thoroughly, and come with solutions — not just questions.
• Collaborative: I adapt quickly to feedback, communicate concisely, and only sync when it actually adds value.
• Proactive: I take initiative, propose improvements, and push projects forward without waiting to be asked.
• Clarity first: I surface the core need, make assumptions explicit, and write code that’s easy to read and maintain.

Beyond work

I like hard problems and steep learning curves — from AI research to climbing Mont Blanc. I’ve consulted for the Ynternet.org Foundation, volunteered as a chef in a student‑run restaurant in Norway, and trained teams in collective intelligence methods for better collaboration.

In my free time, I like reading about the future of technology and politics, but I also enjoy cooking, dancing, and mountain sports.