Curriculum vitae · Amsterdam
Pedro M. P. Curvo
PhD Researcher in Artificial Intelligence, University of Amsterdam
One thing I eventually learned in life is that sometimes you need to brag a little about yourself. To be honest, I hate bragging or making what I am or what I've done sound more complicated than it is. At the same time, you should be able to look at where you are now and be proud of it. So, here goes:
Summary
Hello, everyone! I'm Pedro. I was born and raised in Portugal, in a village close to Spain. I moved to Lisbon when I was 17 and completed two bachelor's degrees, in Engineering Physics and Computer Science. I then moved to the U.S. to work on fusion reactors, where I collaborated with people from Madison and Princeton. The work I did was later used by Proxima Fusion, which you may know as the biggest startup in the field. Eventually, I moved to Amsterdam, where I worked on agents and published that work as a sole author at a NeurIPS workshop. I've also worked at a fintech unicorn handling between six and nine trillion transactions per year, where I developed an agentic framework for its anti-money-laundering system. Over the past year, I've worked on aerodynamics simulations, which were highlighted at CVPR, as well as generative models—specifically, how we can generate something from only a few new examples. For now, I'm a PhD candidate focusing on World Models.
Outside of research and work, I've enjoyed classical music since I was a kid; I guess five years of piano had some influence. I enjoy travelling, seeing new places, and, more importantly, trying new food. I must say I'm lucky because my girlfriend is as much of a foodie as I am, so we can always try twice as many dishes. On the slightly weird hobby side: I like assembling watches. I find the whole process—from ordering the parts to assembling them—somehow relaxing, even though it is arguably the opposite. I guess a new, controllable kind of stress is what it takes to make you forget the others for a while.
I like reading, and I don't love exercising, even though you can find me doing it daily for the sake of my health.
This is a summary of what I've done so far. This text was written at the McDonald's in Schiphol Airport with my friend Lohit, just before we travelled to the introduction days for The Bridge in London. We both hate bragging about ourselves, but if we were going to play the same game as all the braggers we saw in the WhatsApp chat, we needed a good self-introduction—and so we wrote one in a few minutes at the airport. Everything in our introductions is true and is not meant to pass off lies just to look cool or unreachable. In the end, I still believe that great work speaks for itself.
For the PDF version of my CV, download it here . Otherwise, keep reading for its translation to the website below.
A short version
Timeline
- PhD in Artificial IntelligenceUniversity of Amsterdam
- MSc Artificial IntelligenceUniversity of Amsterdam
- BSc Physics EngineeringInstituto Superior Técnico · Lisbon
Education
PhD in Artificial Intelligence
University of Amsterdam — Amsterdam, Netherlands
- Research focus: World Models
MSc Artificial Intelligence (cum laude)
University of Amsterdam — Amsterdam, Netherlands
- GPA: 9/10 (Excellent, A+ in US/UK scale)
- Thesis title: “Follow the Mean: Reference-Guided Flow Matching”
BSc Physics Engineering
Instituto Superior Técnico — Lisbon, Portugal
- Thesis title: “Using Deep Learning to Design High Aspect Ratio Fusion Devices” (20/20)
Publications
- Curvo, P.M.P., Zhdanov, M., Eijkelboom, F., & Van De Meent, J.-W. (2026). Follow the Mean: Reference-Guided Flow Matching. ArXiv, abs/2605.10302.
- Curvo, P.M.P., Van De Meent, J.-W. & Zhdanov, M. (2025). MSPT: Efficient Large-Scale Physical Modeling via Parallelized Multi-Scale Attention. CVPR 2026 (Highlight).
- Curvo, P.M.P. (2025). The Traitors: Deception and Trust in Multi-Agent Language Model Simulations. NeurIPS 2025 Workshop on Multi-Turn Interactions in Large Language Models.
- Curvo, P.M.P., Dragomir, M., Torpes, S., & Rahimi, M. (2025). Reproducibility Study of “Cooperate or Collapse: Emergence of Sustainable Cooperation in a Society of LLM Agents”. ArXiv, abs/2505.09289.
- Curvo, P., Ferreira, D.R., & Jorge, R. (2024). Using deep learning to design high aspect ratio fusion devices. Journal of Plasma Physics, 91.
Work Experience
Feedzai
Software Engineer Intern, AI & Financial Crime — Lisbon, Portugal
- Implemented an agentic framework to automatically draft SAR narratives, reducing analyst drafting time by 70%.
Copernica
Junior Software Engineer — Amsterdam, Netherlands
- Built backend features for an email marketing platform using TypeScript and PostgreSQL.
Teaching Experience
Teaching Assistant
Department of Computer Science, University of Amsterdam — Amsterdam, Netherlands
- Deep Learning I, Information Retrieval I, Machine Learning I, Interpretability and Explainability in AI.
Skills
ML Research
PyTorch, JAX, Diffusers, Transformers, Scikit-Learn
Programming
Python, TypeScript, C++, C
Systems & Tools
Git, Docker, Kubernetes, FastAPI, PostgreSQL, GitHub Actions
Languages
Portuguese (Native), English (Full Professional Proficiency), Spanish (Intermediate)
Last updated: 2026