David González-Martínez

I’m David González-Martínez, an ML researcher and M.Sc. student at the University of Tübingen. I’m currently a student research assistant at the Max Planck Institute for Intelligent Systems and ELLIS Institute Tübingen, working at the WEI Lab under the supervision of Dr. Shiwei Liu. I previously completed my B.Sc. at the University of Seville, with a thesis on model compression.

I am broadly interested in questions that could help make machine learning better. In particular, I think about two axes: making models do more, and making them use fewer resources.

Some recent and current directions I have been thinking about and exploring, among others, are better deep learning architectures and learning paradigms, reinforcement learning/on-policy learning and LLM reasoning, model efficiency and compression, and the science of deep learning.

In the past, I spent some time thinking about ML systems and engineering. A project I spent quite some time on is pequegrad, a toy deep learning framework built from scratch. It features a graph tracing mechanism, which is then used for automatic differentiation, higher-order differentiation, JIT compilation (graph optimizations, pattern matching, and fused CUDA code generation), among other experimental features.

I am always happy to chat about research, especially if it touches any of the questions above.

Featured publications

@misc{gonzalez-martinez2025balf,
  title        = {BALF: Budgeted Activation-Aware Low-Rank Factorization for Fine-Tuning-Free Model Compression},
  author       = {David González-Martínez},
  year         = {2025},
  eprint       = {2509.25136},
  archivePrefix= {arXiv},
  primaryClass = {cs.LG},
  doi          = {10.48550/arXiv.2509.25136}
}

For a complete list of publications, please see my publications page.