====== Michael Schaarschmidt ======
**Michael Schaarschmidt** is a research leader at Isomorphic Labs focused on foundational machine learning research for computational biology and drug discovery. His work centers on developing and refining core ML techniques that underpin protein structure prediction and molecular design applications(([[https://www.theneurondaily.com/p/watch-how-isomorphic-labs-works-to-drug-undruggable-diseases|The Neuron - "Watch How Isomorphic Labs Works to Drug Undruggable Diseases" (2026]])).

===== Research Focus =====
Schaarschmidt leads the foundational AI research division at Isomorphic Labs, an organization established as a sister company to DeepMind to advance AI-driven drug discovery. His research program addresses critical modeling challenges in computational biology that often remain unpublished or underspecified in academic literature. This includes investigating the implementation details, parameter configurations, and architectural innovations necessary to translate theoretical advances in protein structure prediction into practical molecular design systems(([[https://www.theneurondaily.com/p/watch-how-isomorphic-labs-works-to-drug-undruggable-diseases|The Neuron - "Watch How Isomorphic Labs Works to Drug Undruggable Diseases" (2026]])).

The foundational research conducted under his leadership encompasses the mathematical foundations and optimization techniques required for accurate structural biology modeling. This work supports the broader [[isomorphic_labs|Isomorphic Labs]] mission to identify and develop treatments for diseases traditionally considered undruggable—conditions where conventional pharmaceutical approaches have proven insufficient or ineffective.

===== Protein Structure and Molecular Design =====
A central focus of Schaarschmidt's research involves refining machine learning approaches to protein structure prediction, building upon advances like AlphaFold and related architectures. Beyond structure prediction, his team works on extending these capabilities toward functional molecular design—enabling AI systems to not only predict how proteins fold, but to design novel molecular entities with desired properties for therapeutic applications(([[https://www.theneurondaily.com/p/watch-how-isomorphic-labs-works-to-drug-undruggable-diseases|The Neuron - "Watch How Isomorphic Labs Works to Drug Undruggable Diseases" (2026]])).

The research addresses practical challenges in structural biology that academic publications may not fully document, such as handling edge cases in protein folding predictions, managing computational constraints, and optimizing model performance across diverse molecular families. This commitment to unpublished implementation details reflects the empirical demands of working on real pharmaceutical problems where theoretical understanding must be complemented by careful engineering.

===== Isomorphic Labs Context =====
Isomorphic Labs, founded by DeepMind as a dedicated drug discovery company, represents a significant application of machine learning to one of science's most challenging domains. Schaarschmidt's leadership of the foundational research ensures that the underlying ML techniques remain at the forefront of scientific capability. The organization focuses specifically on leveraging AI to tackle molecular diseases and drug targets that have resisted conventional approaches, making precision in core ML techniques essential to the mission's success.

The separation of foundational research from applied drug discovery within Isomorphic Labs' structure allows Schaarschmidt's team to systematically investigate the fundamental modeling challenges that enable the company's therapeutic programs. This includes contributing to publications in structural biology and machine learning while maintaining the proprietary insights necessary for competitive advantage in drug discovery applications.


===== See Also =====
  * [[isomorphic_labs|Isomorphic Labs]]
  * [[jurgen_schmidhuber|Jürgen Schmidhuber]]
  * [[andrej_karpathy|Andrej Karpathy]]
  * [[ai_first_drug_discovery|AI-First Drug Discovery]]

===== References =====