I am a PhD student at Stanford University. My dream is to design the next generation of learning-based assistive robots. I plan to get there by studying methods for in-context adaptation of policies to unseen scenarios while emphasizing human-centric approaches.

Some of my most renowned research includes the first Real-to-Sim-to-Real work and designing the first memory VLA at Physical Intelligence with MEM, which we then brought into π_0.7.

Previously, I obtained my Master's degree at Harvard University supported by the Real Colegio Complutense scholarship. And I was part of the Improbable AI Lab at MIT CSAIL under the supervision of Professors Pulkit Agrawal and Abhishek Gupta. Even before that, I received my Bachelor's degree from École Polytechnique Fédérale de Lausanne (EPFL), where I was lucky enough to do my bachelor's project at MLO and iGH under the supervision of Professors Martin Jaggi and Mary-Anne Hartley. There I started an open source platform, DisCO, for training ML models in a federated and decentralized fashion. I showed its usefulness training Computer Vision models for Healthcare.

In industry, I have interned at Physical Intelligence where I worked on adding memory to VLAs. Before that, I interned at Microsoft Bing and Amazon Search working on their ranking algorithms.

I spend my free time running, sailing, and biking; playing the baroque guitar; or reading books.

My dream is to give people the tools to dedicate their life to what they really want to do. If you had to dream big, what would you do?

How do I plan to do this? I want to develop robots that expand human capabilities. Robots that will help us build new amazing physical things, from helping increase productivity in agriculture, construction and manufacturing, to exoskeletons that will help regain the mobility of those impaired, to a robot that will help you with your hobbies like building your next watch or windsurf board.

Giving such robots to the world will expand human capabilities and resources to the point that I hope they will give economic independence to the general population and allow people to dedicate their life to whatever they want.

I dedicate 99% of my professional energy on the above. But I acknowledge that getting to do what you really want can also be bottlenecked by lack of experience or guidance. Many people have guided and helped me get where I am, my dream school and dream lab to do the research I want to do. I try spending some time helping others who, like me, were uncertain how to reach their Moonshot goals, through Project Moonshot.

News

---

• Oct 2026 Speaker at the 3rd EFCL International Workshop on Sustainable AI (ETH Zurich), presenting “Essential Ingredients for Learning Long-Horizon Robot Policies.” Video
• Sep 2026 Three papers accepted at CoRL 2026: Freeform Preference Learning, MEM (Spotlight, top 3.4%), and π_0.7 (Oral, top 1%).
• Jul 2026 Presenting our work at the Y Combinator Paper Club (Jul 29). Video
• Jul 2026 Speaker at the ExWBC Workshop at RSS 2026 (Sydney, Jul 13). Video
• Jul 2026 π^*_0.6 presented at RSS 2026.
• 2025–26 Interned at Physical Intelligence (Jun 2025 – Mar 2026), where I worked on adding memory to VLAs.
• Dec 2025 Robot Olympics at Physical Intelligence: I was responsible for all of the cleaning tasks, which used the first iterations of the MEM video encoder I was working on.
• Oct 2025 Invited talk on Real-to-Sim-to-Real visuomotor robot learning at an IROS 2025 workshop. Video
• Jun 2025 Best Paper Award at the Robot Representations Workshop, RSS 2025, for Learning Long-Context Diffusion Policies.
• Sep 2024 Started my PhD at Stanford University, advised by Prof. Chelsea Finn.