I am very excited to announce that our latest work "Thermodynamics-informed Graph Neural Networks" has been accepted for publication in IEEE Transactions on Artificial Intelligence! #IEEE🎉​​😄​ In this work we present a deep learning simulator for dissipative dynamical systems.

I'll also be presenting a poster on my latest work at @Penn on coarse-graining models at the "Machine Learning and the Physical Sciences" Workshop (Saturday 6th). See you there! ⚛️ #AI4Simulation #NeurIPS2025 #CoarseGraining #Physics

Next week I'll be at @NeurIPSConf in San Diego (Dec 1-7)! Hmu if you wanna grab a coffee and chat about #AI4Science, or come by the Emmi AI booth to see the cool stuff we're building! 🚀💫

#deeplearning #machinelearning #thermodynamics #physics #simulation #polymers #colloids #computationalphysics #AI #coarsegraining #materials

📝 Read the preprint: https://t.co/alve2hXE9a 💻 Explore the code:

We provide open-source implementations, enabling large-scale inference and extensibility to diverse particle systems. This project was developed during my postdoc at @Penn, together with the amazing collaborators Max Win, Thomas C. O'Connor, Paulo E. Arratia, and Nat Trask. 👏👏

🔬 Applications we showcase: 1️⃣ Coarse-graining star polymers at challenging levels while preserving non-equilibrium statistics 2️⃣ Learning from high-speed videos of colloidal suspensions to capture local rearrangements and emergent stochastic dynamics

We further introduce a self-supervised strategy to identify emergent structural variables, making the approach broadly applicable.

We propose a metriplectic ML framework that learns coarse-grained dynamics directly from trajectory data while: ✅ preserving the first and second laws of thermodynamics ✅ conserving momentum ✅ enforcing a fluctuation–dissipation balance to capture non-equilibrium statistics

Multiscale systems are everywhere in science, but simulating them efficiently remains a huge challenge. When high-dim particle dynamics are coarse-grained into lower-dim models, important information is lost leading to dissipative, history-dependent, and stochastic physics.

I'm very excited to announce that our latest work "Data-driven particle dynamics: Structure-preserving coarse-graining for emergent behavior in non-equilibrium systems" is now available online! 🥳🥳

Very proud to announce that I've joined Emmi AI as a Senior Research Engineer! 🎉🎊 I'll be working on applying physics-based AI models into industry scale problems. Very thankful to @dennisjust, Miks Mikelsons and @jo_brandstetter for the opportunity! #AI4Simulation #EmmiAI

Visiting beautiful Paris for the #AICOMAS2025 Conference. Make sure you don't miss my first keynote of my recent work @Penn, about coarse-graining of MD models! See you on Wednesday! 😄

It was incredible to see all my international colleagues again and share new ideas and thoughts about science. See you soon! 😄😄 #PhDOlympiad #ECCOMAS2024 #AI4Science #Simulation

I'm so excited to share that I've won the @ECCOMASofficial PhD Olympiad 2024! Many thanks to the organizers Konrad Perzyński, Paulina Stempin and of course my thesis advisors @EliasCueto, @a_badias for making all this possible. 🎉🎊

. @QuercusEtAl defending his candidacy to the european PhD Olympiad of #eccomas2024. Well done, Quercus!

On my way to Lisbon! Tomorrow I will be presenting a glimpse of my thesis "Structure-Preserving Deep Learning" in the 14th PhD @ECCOMASofficial Olympiad, representing @semni_social. See you there! 😄 #ECCOMAS2024 #PhDOlympiad #AI4Science

.@unizar recibe 1,6M€ para investigar en inteligencia artificial en #automoción y #construcción. Investigadores del @I3Aunizar liderarán la Cátedra ENIA, concedida por @mintradigital junto a la empresa @ESIgroup y @ArtsetMetiers_ de París.

Great news! 🥳 I've been awarded with the best 2023 thesis prize from @semni_social! Many thanks to my advisors @EliasCueto, @a_badias and everyone that, in one way or another, contributed to my thesis! 😄 📝 Check it here: https://t.co/GEQckN1N2q #PhDAward #AI4Science #ML

I'm thrilled to share that I've started a Postdoc position @Penn! 🎉🎊 I'll be working with Nat Trask on scientific deep learning and structure-preserving algorithms. Looking forward to learning and contributing to this amazing field with new ideas! #AI4Science #SciML #UPennMEAM

Last Friday, I successfully defended my PhD thesis! 🥳Many thanks to the committee (OlivierAllix, @AldakheelFadi, IciarAlfaro), advisors (@EliasCueto, @a_badias), friends and family, I had such a great time! Excited to begin the next chapter of my career. Stay tuned...