Robert Pinsler

Robert Pinsler is a machine learning researcher whose work connects robot learning, active learning, molecular design and AI for materials science. He earned a BSc in Business Informatics at DHBW Mannheim in 2014 and an M.Sc. in Computer Science at the Technical University of Darmstadt in 2017 before joining Cambridge's Machine Learning Group as a PhD student in October 2017. His papers covered human-feedback reinforcement learning, Bayesian batch active learning and symmetry-aware molecular design. He coauthored MatterGen in December 2023; Nature published its paper in January 2025, and a May 2026 report described synthesis of MatterSim-v1-screened TaP. As of October 2026, he is a Senior Researcher at Microsoft Research Cambridge, working on machine learning for materials discovery.

Now at: Microsoft Research
Works: 17
Citations: 947
h-index: 9
Badges: Bridge

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2014–2017

Darmstadt master's and Cambridge PhD

He earns a BSc in Business Informatics at DHBW Mannheim in 2014. In 2017, he earns an M.Sc. in Computer Science from the Technical University of Darmstadt, working with Gerhard Neumann, Jan Peters and Stefan Roth. In October 2017, he joins the University of Cambridge's Machine Learning Group as a PhD student under Carl Rasmussen.
2018

Human feedback for robot learning

In 2018, he coauthors *Sample and Feedback Efficient Hierarchical Reinforcement Learning from Human Preferences*, which learns reward functions from robot and human perspectives to improve sample and feedback efficiency.
2019

Bayesian batch active learning

In 2019, he coauthors *Bayesian Batch Active Learning as Sparse Subset Approximation*, which creates diverse batches for efficient active learning at scale.
June 2020–September 2020

Research internship at Amazon

From June 2020 to September 2020, he works as a research intern with Amazon's AWS AI Platform in Berlin on data-efficient meta-learning.
May 2021

Symmetry-aware molecular design

In May 2021, he coauthors *Symmetry-Aware Actor-Critic for 3D Molecular Design*. The deep reinforcement learning method uses a spherical harmonics representation to exploit symmetry and generate 3D structures unavailable to earlier methods.
December 2023–January 2025

MatterGen for materials design

In December 2023, he coauthors MatterGen, a diffusion model for stable materials with adapter modules that can be fine-tuned to generate materials under chemistry, symmetry and property constraints. Nature publishes its paper in January 2025 and lists him among seven authors who contributed equally. MatterGen generates stable, diverse inorganic materials across the periodic table and can be fine-tuned for property constraints.
May 2026

MatterSim identifies TaP

In May 2026, he coauthors a Microsoft Research report on screening more than 240,000 candidate materials with MatterSim-v1. The team identifies tetragonal tantalum phosphorus (TaP) as a potential high thermal conductor and reports its experimental synthesis, with measured conductivity close to silicon's.
As of October 2026

Senior researcher at Microsoft

As of October 2026, he is a Senior Researcher at Microsoft Research Cambridge, where he works on machine learning for materials discovery.

Career

Since 2023Microsoft ResearchSenior Researcher
2020AmazonResearch Intern

Education

2017TU DarmstadtM.Sc. in Computer Science
From 2017University of CambridgePhD student
2014Baden-Wuerttemberg Cooperative State University MannheimBSc in Business Informatics