John Jumper is an AI researcher and biophysicist who co-led AlphaFold2 and led AlphaFold3 development at Google DeepMind. He shared the 2024 Nobel Prize in Chemistry for protein-structure prediction. In June 2026, he announced that he would leave Google DeepMind to join Anthropic after a break.[1][9][2][3][4]
Start with AlphaFold: Five years of impact →
Education
Links
Royal Society biography (historical title)
X University of Chicago biography (historical title) ORCID Background
Jumper graduated from Vanderbilt in 2007 with training in physics and mathematics. He pursued physics at Cambridge and completed an MPhil in theoretical condensed matter physics in 2008. He then worked at D. E. Shaw Research on molecular dynamics, computer simulations of how proteins move. That experience drew him toward biological problems and a chemistry PhD at the University of Chicago. Chicago identifies him as an SM graduate of 2012 and PhD graduate of 2017.[5][6][7][1]
Career
Contributions
Learning to predict a protein’s structure
A protein’s amino-acid sequence is relatively straightforward to determine compared with its three-dimensional structure. AlphaFold2 uses neural networks to predict that structure, learning from biological sequence information and known structures. The 2021 paper reported strong results in the CASP14 assessment. It credits Jumper and Demis Hassabis with leading the research and Jumper with work on the architecture, training and end-to-end approach. The result helps researchers investigate proteins, but a predicted structure is a model with uncertainty, not a direct experimental observation.[9][1]
From individual proteins to molecular complexes
AlphaFold3 broadened the task to structures containing combinations of proteins, DNA, RNA, small molecules and other chemical components. Its 2024 paper describes a diffusion-based architecture for predicting those assemblies together. This matters because biological activity often depends on molecules interacting, rather than on an isolated protein’s shape. Jumper led the system’s development with the research team. The model’s scope should not be mistaken for a guarantee about every target, molecular motion or whether a proposed drug will work in a patient.[10][1]
Explaining AlphaFold’s scientific impact
The co-authored Five years of impact article is an accessible starting point because it explains proteins, their structures and the practical research questions AlphaFold addresses before discussing the models. It also links to a conversation with Jumper about the development of AlphaFold2 and AlphaFold3. Follow that with the original papers to understand what was evaluated and how confidence is reported. The distinction between a useful structural hypothesis and a measured biological result is central to reading this work accurately.[11][9][10]
Tools, writing & talks
Writing, research & talks
Start here
AlphaFold: Five years of impactdeepmind.googleA co-authored explanation of AlphaFold’s scientific purpose and research use, with an accompanying conversation featuring Jumper. This is the most approachable first stop before the architecture papers.
Introduces the scientific problem and practical use of AlphaFold, with a route to Jumper’s own explanation and the technical papers.
Co-author on behalf of the AlphaFold team[11]The AlphaFold2 methods paper explains the model, its evaluation and confidence estimates. Jumper is credited with research leadership and architecture development; the author contribution statement makes the team’s work explicit.
Co-author and research co-lead[9]The AlphaFold3 paper describes a model for predicting molecular complexes involving proteins, nucleic acids and small molecules. It expands the prediction task while also discussing limitations and uncertainty.