I’m Pau Hidalgo Pujol, an artificial-intelligence student and researcher from Banyoles, based in Barcelona. I have worked at Nostrum Biodiscovery on geometric deep learning for protein structures.

I studied Artificial Intelligence at the Universitat Politècnica de Catalunya. During a 2025–2026 exchange at ETH Zürich, I took master-level courses in probabilistic AI, neural-network theory, reliable and trustworthy AI, and learning in biological and artificial neural networks.

Research interests

My main interests fall into three areas:

  1. Mathematical foundations of machine learning: probability, optimisation, learning theory, generalisation, invariance, and representation learning.
  2. Geometric machine learning for biology: graph neural networks, protein representation learning, graph construction, sparsification, and efficient scientific ML.
  3. Music, sound, and machine learning: audio representations, symbolic music, generation, and computational creativity.

My bachelor’s thesis, completed with Telefónica I+D and UPC, studied whether eye-tracking signals can help train, evaluate, and interpret language models. Other projects include Cucafera, a Catalan language model trained from scratch, and Patufet, its public dataset collection.

Music

I studied cello at the Conservatori de Música de Girona and received a professional music diploma in 2023.

Curriculum vitae

You can view my CV here or download the full CV as a PDF.

Contact

Email is the most direct way to reach me: pauhidalgopujol@gmail.com.

I also publish code on GitHub, models and datasets on Hugging Face, and professional updates on LinkedIn.