Matthew Horton
Also known as Matthew K. Horton, Matthew Kristofer Horton, Matt Horton
Matthew Horton is a materials scientist and software engineer who helped create MatterGen, a generative AI model for designing inorganic materials with target properties. After a PhD at Imperial College London on dislocations in semiconductors for LED lighting, he moved to Berkeley in 2016 on a Lindemann Fellowship. At Lawrence Berkeley National Laboratory, he helped lead the Materials Project and was first author of Crystal Toolkit, a framework for materials science web apps. He later joined Microsoft Research AI for Science, where he was one of seven equal-contribution authors of MatterGen, published in Nature in January 2025; in May 2025, he joined the founding team of Periodic Labs. He now consults through Isotropy, LLC.
Timeline
-
By 2016
From Imperial to Berkeley
His PhD at Imperial College London studies dislocations in semiconductors for more energy-efficient LED lighting, and an EPSRC Doctoral Prize award lets him continue the work there. In 2016, he takes up a Lindemann Fellowship at Berkeley and researches solar energy materials at Lawrence Berkeley National Laboratory.
-
2019
Predicting magnetic order at scale
In Kristin Persson's research group at Lawrence Berkeley National Laboratory, he works on magnetocalorics and high-throughput materials discovery. In June 2019, he is first author of an automated workflow, built with the atomate code, that predicts the magnetic ground state of inorganic crystals; it correctly predicts a non-ferromagnetic ground state in 95% of 64 benchmark materials.
-
By 2020
Helping lead the Materials Project
As a staff developer at Lawrence Berkeley National Laboratory, he helps lead the Materials Project, an open, academic effort to share high-quality materials data, and he also helps create matsci.org, a shared space for the computational materials science community. In September 2020, he gives a Missouri State University seminar titled "Democratizing Access to Materials Science with the Materials Project".
-
February 2023
The Crystal Toolkit framework
In February 2023, he is first author, with 23 coauthors, of Crystal Toolkit, an open-source framework for building web apps that view, analyze and transform crystal structures and other materials data. Crystal Toolkit powers the Materials Project website frontend.
-
December 2023–January 2025
MatterGen at Microsoft Research
After leaving Persson's group, he joins Microsoft Research AI for Science in Redmond, where he is a Senior Research SDE when the team announces MatterGen, a generative model for property-guided materials design, in December 2023. Nature publishes MatterGen in January 2025 with him among seven authors who contributed equally. Its structures are more than twice as likely to be new and stable as those of earlier generative models, and the team synthesizes one generated material whose measured property lands within 20% of its target.
-
May 2025
Founding team at Periodic Labs
In May 2025, he leaves Microsoft Research AI for Science and joins Periodic Labs as a founding team member, part of the startup's eight-person stealth team. The company is announced publicly in September 2025.
-
July 2025
A Materials Project perspective
In July 2025, he is first author of a Nature Materials Perspective on the Materials Project, which launched formally in 2011 and is used by more than 600,000 materials researchers. The article describes how the project's data platform and software ecosystem have shaped data-driven materials science.
-
As of October 2026
Consulting through Isotropy
He is a materials scientist and software engineer who consults through Isotropy, LLC. His former employers include Periodic Labs, Microsoft Research and Berkeley Lab.
Career
- NowIsotropy, LLCConsultant
- From 2025Periodic LabsFounding team member
- 2023–2025MicrosoftSenior Research SDE
- 2014–2025Office of ScienceStaff developer
- 2014–2025University of California, Berkeley
Education
Most-cited works
- A generative model for inorganic materials designWith Aliaksandra Shysheya, Andrew Fowler, Claudio Zeni, Daniel Zügner, Robert Pinsler and Ryota Tomioka
- Atomate: A high-level interface to generate, execute, and analyze computational materials science workflows
- Accelerated data-driven materials science with the Materials Project
- Grain boundary properties of elemental metals
- High-throughput predictions of metal–organic framework electronic properties: theoretical challenges, graph neural networks, and data exploration
Sources and credits
Photo: We have no published photo of Matthew Horton yet. Send us a free-licensed photo.
- Physics - Matthew K. Horton
- High-throughput prediction of the ground-state collinear magnetic order of inorganic materials using Density Functional Theory | npj Computational Materials
- Kristin Persson's Research Group: People
- mkhorton (Matt Horton) · GitHub
- PAMS Seminar: Dr. Matthew Horton "Democratizing Access to Materials Science with the Materials Project" - Calendar of Events - Missouri State University
- [2302.06147] Crystal Toolkit: A Web App Framework to Improve Usability and Accessibility of Materials Science Research Algorithms
- MatterGen: Property-guided materials design - Microsoft Research
- A generative model for inorganic materials design | Nature
- MatterGen: A new paradigm of materials design with generative AI - Microsoft Research
- Periodic Labs — Hiring Analysis
- Accelerated data-driven materials science with the Materials Project | Nature Materials
- Matthew Horton



