Miguel Arbesú Andrés

Miguel Arbesú Andrés

Researcher in Bio ∩ AI

InstaDeep Ltd., Berlin, Germany

Hello

My name is Miguel. I am a researcher working at the intersection between Biology and Artificial Intelligence.

Nowadays I am interested in AI-powered protein engineering and optimization problems. I specialize in designing flexible proteins that challenge structure-based modeling.

Interests
  • Protein Engineering
  • Optimization
  • Immunology
Education
  • PhD in Organic Chemistry, 2018

    University of Barcelona, Spain

  • MSc in Organic Chemistry, 2013

    University of Barcelona, Spain

  • BSc, MSc in Chemistry, 2012

    University of Oviedo, Spain

Experience

 
 
 
 
 
InstaDeep Ltd.
Senior applied research scientist
Sep 2024 – Present Berlin, Germany
Develop and apply state-of-the-art methods in protein engineering and other optimization problems.
 
 
 
 
 
InstaDeep Ltd.
Research engineer
Feb 2023 – Aug 2024 Berlin, Germany
 
 
 
 
 
Max-Delbrück-Centrum für Molekulare Medizin (MDC)
Visiting researcher
Sep 2022 – Nov 2022 Berlin, Germany

Helmholtz Information & Data Science Academy (HIDA) grantee at the MDC Bioinformatics unit.

Projects:

  • Protein:drug interaction prediction with geometric deep learning and protein language models.
 
 
 
 
 
Leibniz-Forschungsinstitut für Molekulare Pharmakologie (FMP-Berlin)
Postdoctoral researcher
Mar 2018 – Jun 2022 Berlin, Germany

Projects:

  • Liquid-liquid phase separation of human FUS by solid state Nuclear Magnetic Resonance .
  • Regulation of plant salt-stress response by phosphorylation of protein CC1.
  • Structural disorder in enteropathogenic bacterial proteomes.

Tasks:

  • Method development
  • Solid state and solution Nuclear Magnetic Resonance data acquisition
  • Data analysis
 
 
 
 
 
BioNMR group - University of Barcelona
PhD student
Sep 2013 – Feb 2018 Barcelona, Spain

Thesis: A novel regulatory unit in the N-terminal region of c-src.

Tasks:

  • Protein expression and purification.
  • Solution Nuclear Magnetic Resonance data acquisition.
  • Data analysis.

Latest publications

(2025). GeoGraph: Geometric and Graph-based Ensemble Descriptors for Intrinsically Disordered Proteins. In Arxiv. Accepted in the NeurIPS 2025 AI for Science and Machine Learning and the Physical Sciences workshops.

PDF Cite Code DOI

(2024). Generative Model for Small Molecules with Latent Space RL Fine-Tuning to Protein Targets. In Arxiv.

PDF Cite DOI

(2023). Offline RL for generative design of protein binders. In BioRxiv. Accepted in the NeurIPS 2023 New Frontiers of AI for Drug Discovery and Development workshop.

PDF Cite DOI

Upcoming & Past Talks

Foundational Models for Genomics: Decoding Genomic Sequences
Workshop hosted by BLISS.
Foundational Models for Genomics: Decoding Genomic Sequences

Contact