Eric Alcaide
I'm a researcher working on LLMs, coding agents, and AI4Science.
My current research interests focus on:
- Efficient processing of information with Recurrent Neural Networks
- Adaptive computation & reasoning
- Agents and LLMs for code
- AI for Science and BioTech (3D, design, automation)
Previously, I was one of the pioneers of CoFolding early in 2021 as the first member of CHARM Tx, and worked on ML for structural biology at Vir Biotechnology. I am also an open source researcher at EleutherAI, where I led efforts on Linear Attention (author of RWKV) and AI for Science - OpenBioML community.
I hold a Physics degree and a Medicine degree (MD) from the University of Barcelona. My medical thesis, "Towards immune profiling of brain metastasis from lung cancer as a driver for precision medicine", was supervised by Ernest Nadal Alforja, MD, PhD at the ICO.
ericalcaide1 [at] gmail [dot] com
News
- Jan 2026- Multiple Token Divergence accepted at ICLR 2026 — See you in Rio!
- Dec 2025- Around for NeurIPS2025 in San Diego. Let's chat!
Selected Publications
Multiple Token Divergence: Measuring and Steering In-Context Computation Density