I’m a research scientist working on the optimization of large personalized models.
At Spotify, I help build learning systems that understand musical taste and shape how people discover and connect with music. To tackle the challenges of learning at Spotify scale—roughly 10% of the world’s population generating 3.4 trillion taste signals each day—my research brings together optimization and data-centric approaches to model training. I focus on targeted data selection, continual learning, and training recipes that scale effectively with data and compute while accounting for business and resource constraints, as well as requirements on model behavior.
More broadly, I’m interested in the understanding and holistic design of learning systems as models grow more capable, datasets reach unprecedented scales, and tools, harnesses, and agentic workflows add new layers of complexity.
Selected Publications
See also my Google Scholar profile.
Near-Optimal Solutions of Constrained Learning Problems
Juan Elenter, Luiz Chamon, Alejandro Ribeiro
International Conference on Learning Representations (ICLR), 2024
From Habits to Discovery: Deploying LLMs for Personalized Generative Recommendations at Spotify
Edoardo D'Amico, ..., Juan Elenter, ..., Paul N. Bennett
ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2026
TSVD: Bridging Theory and Practice in Continual Learning with Pre-trained Models
Liangzu Peng, Juan Elenter, Joshua Agterberg, Alejandro Ribeiro, René Vidal
International Conference on Learning Representations (ICLR), 2025
Efficient Dataset Selection for Continual Adaptation of Generative Recommenders
Cathy Jiao, Juan Elenter, Praveen Ravichandran, et al.
CAO Workshop at ICLR 2026 (Oral)
A Lagrangian Duality Approach to Active Learning
Juan Elenter, Navid NaderiAlizadeh, Alejandro Ribeiro
Conference on Neural Information Processing Systems (NeurIPS), 2022
Feasible Learning
J. Ramirez * , I. Hounie * , J. Elenter *, J. Gallego *, A. Ribeiro, S. L. Julien
International Conference on Artificial Intelligence and Statistics (AISTATS), 2025
Neural Networks with Quantization Constraints.
Ignacio Hounie*, Juan Elenter*, Alejandro Ribeiro
IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023
A little more about me
I’m from Montevideo, Uruguay, where I studied engineering, focusing on signal processing and machine learning. I also interned at CERN, building visualization tools for large particle collision datasets in the CMS Open Data initiative. I later completed master’s degrees in statistics and systems engineering at UPenn and was a visiting student at Stanford’s Kundaje Lab, where I worked on BasepairModels, a Python library for regulatory genomics analysis.
I’m also a music nerd who enjoys sailing. In 2016, I sailed across the Atlantic with my father, Pablo.
Leadership and community
- Between 2011 and 2015, I played for Uruguay's national water polo team.
- I co-organized the NeurIPS 2025 Workshop on Constrained Optimization for Machine Learning to discuss progress in building learning systems that excel at their primary task while satisfying strict requirements.
- I serve on the Junior Board of ReachingU, which has invested more than $19 million to expand educational opportunities in Uruguay.