John Jumper

Type: collaborator Slug: collaborator—john-jumper Sources: highly-accurate-protein-structure-prediction-with---hassabis, protein-structure-predictions-to-atomic-accuracy-with-alphafold—hassabis, applying-and-improving-alphafold-at-casp14—hassabis, highly-accurate-protein-structure-prediction-for-the-human-proteome—hassabis Last updated: 2026-05-13


Summary

John Jumper is the first author and lead architect of AlphaFold2, appearing on 4 corpus papers (2020–2022). He conceived and implemented the Evoformer architecture that achieved atomic-accuracy protein structure prediction. He shared the 2024 Nobel Prize in Chemistry with Demis Hassabis.

Role

First author on the landmark Nature paper (paper—highly-accurate-protein-structure-prediction-with---hassabis). While Hassabis provided strategic leadership and the broader AI-for-science vision, Jumper was the hands-on architect who designed the Evoformer and structure module. The Nobel Prize recognised both as equal contributors.

Papers together

  1. Improved protein structure prediction using potentials from deep learning (2020, precursor)
  2. Highly accurate protein structure prediction with AlphaFold (2021, landmark)
  3. Applying and improving AlphaFold at CASP14 (2021)
  4. Highly accurate protein structure prediction for the human proteome (2022)

Connections

  • Project: project—AlphaFold2
  • Themes: theme—protein-folding, theme—AI-for-science
  • Period: period—alphafold-era
  • Claim: claim—learnable-nature-conjecture (AlphaFold2 is the primary evidence)

Honest Gaps

  • No interview or first-person account from Jumper in the corpus — the Jumper-Hassabis intellectual relationship is entirely inferred.
  • Jumper’s background (theoretical physics PhD at Chicago, postdoc at DeepMind) is not documented in any corpus source.
  • The specific division of labour between Jumper’s architecture work and the broader AlphaFold team is unclear.
  • Jumper does not appear on the AlphaFold-Multimer paper — his specific role in the extended AlphaFold programme is undetermined.