Irina Espejo Morales
Irina Espejo Morales is a Research Scientist in AI for Science, working on machine learning for physics, chemistry, and biology.
AI for Science
Irina Espejo Morales
Research Scientist in AI for Science
About
Hi, I’m Irina! I’m a Research Scientist at PolymathicAI and the NYU Center for Data Science, working (very) broadly on AI for Science across physics, chemistry, and biology (catch them all!).
My work has spanned agentic systems for NMR spectroscopy, multimodal foundation models for biomolecular data, and AI for chemistry at IBM Research Zurich. Before that, I completed a PhD in Data Science at NYU with Kyle Cranmer, applying active learning to searches for new physics at the LHC.
My PhD was supported by Fulbright, DeepMind, and IRIS-HEP scholarships. I previously studied theoretical physics at Oxford and earned bachelor’s degrees in Physics and Mathematics from Universitat Autonoma de Barcelona.
Updates
-
-
Preprint is out for MIMIC: A Generative Multimodal Foundation Model for Biomolecules.
Selected Publications
-
NMR Elucidation as an Agentic Search Problem, Not a Modeling Problem
ICML 2026 Workshop AI4Science (Spotlight Talk), 2026
-
MIMIC: A Generative Multimodal Foundation Model for Biomolecules
arXiv preprint, 2026
-
Semantic Search for 100M+ Galaxy Images Using AI-generated Captions
The Astrophysical Journal, 2026
-
Efficient search for new physics using Active Learning in the ATLAS Experiment
Journal of Physics: Conference Series, 2026
-
Scaling MadMiner with a deployment on REANA
Journal of Physics: Conference Series, 2026
-
Unified lookup tables: training foundation models on encoded data
Machine Learning: Science and Technology, 2025
-
Making Sense of Data in the Wild: Data Analysis Automation at Scale
arXiv preprint, 2025
-
Activity recognition in scientific experimentation using multimodal visual encoding
Digital Discovery, 2025
-
Interleaving Text and Number Embeddings to Solve Mathemathics Problems
arXiv preprint, 2024
-
MadMiner: Machine Learning-Based Inference for Particle Physics
Computing and Software for Big Science, 2020
-
Constraining effective field theories with machine learning
EPJ Web of Conferences, 2020
† Equal contribution