Javier Ferrando Monsonís
Bio
Javier Ferrando Monsonís is a researcher in mechanistic interpretability and natural language processing, affiliated with the Barcelona Supercomputing Center (BSC-CNS) and completing a PhD at the Universitat Politècnica de Catalunya (UPC) in the TALP research group. His research focuses on understanding the internal workings of transformer-based language models, with particular interest in their safety implications. He was a MATS (ML Alignment Theory Scholars) scholar mentored by Neel Nanda at Google DeepMind. During his PhD he completed research internships at Meta AI (FAIR), Apple (Machine Translation team), and Amazon (Books Science team), and subsequently joined Amazon as an Applied Scientist. His notable publications include an ICLR 2025 Oral paper on entity recognition and hallucinations in LLMs using sparse autoencoders, a widely-cited primer on the inner workings of transformer language models (2024), and tools for large-scale automated interpretability of language models.
Links
- Personal Website
- https://javiferran.github.io/
- Twitter / X
- LessWrong
- -
Grants
No grants recorded.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.
Details
- Last Updated
- Mar 22, 2026, 4:52 PM UTC
- Created
- Mar 20, 2026, 3:00 AM UTC