David Salinas

David studied theoretical computer science at École Normale Supérieure de Lyon and holds a PhD from the University of Grenoble. He has worked at Amazon and NAVER LABS Europe.

He is currently working on LLMs4EU and OpenEuroLLM, where he leads the evaluation work package. His research focuses on machine learning, particularly in AutoML, LLMs, and predictions for tabular and time-series data. In the past, he also worked on computational topology and geometry processing.

He is very keen to contribute to open-source, he is a core developer of Syne Tune (Hyperparameter Optimization) and SlurmPilot (Python wrapper to schedule experiments on Slurm). He was also a core-developer of GluonTS (time series forecasting), Datawig (data imputation) and Gudhi (computational topology). For more information, you can check David's personal website and his Google Scholar profile.

Selected Publications:

- Tuning LLM Judge Design Decisions for 1/1000 of the Cost (ICML 2025)

- TabArena: A living benchmark for machine learning on tabular data (Neurips DBT 2025 spotlight)

- Syne Tune: A library for large scale hyperparameter tuning and reproducible research (AutoML 2022)

- DeepAR: Probabilistic forecasting with autoregressive recurrent networks (International Journal of Forecasting 2020)