PeopleofSuperintelligence

Researcher

Patricia Suriana

Patricia Suriana is a machine learning researcher who builds deep learning methods for 3D molecular structure, from protein models to small-molecule drug design. With Ron Dror, she coauthored the geometric vector perceptron paper presented at ICLR 2021 and ATOM3D, a benchmark collection that received a NeurIPS 2021 Datasets and Benchmarks Award. She wrote FlexVDW on protein flexibility in ligand docking in 2023 and completed a Stanford computer science PhD in 2024 with a dissertation on equivariant neural networks. Her papers in 2025 and 2026 covered electron-density inputs for 3D learning, language models for small-molecule drug design, and the SynLaD molecule generator. She works at Prescient Design, Genentech.

Now at
Roche
Badges

Timeline

  1. 2020–2021

    Geometric vector perceptrons

    With Bowen Jing, Stephan Eismann, Raphael Townshend and Ron Dror, she coauthors "Learning from Protein Structure with Geometric Vector Perceptrons", posted in 2020 and presented at ICLR 2021. The paper introduces geometric vector perceptrons, which extend standard dense layers to operate on collections of Euclidean vectors, and applies them to model quality assessment and computational protein design.

  2. 2021

    A NeurIPS benchmarks award

    She is one of 13 authors of ATOM3D, a collection of benchmark datasets for 3D molecular learning, first posted in 2020 with Raphael Townshend as first author. The paper appears in the NeurIPS 2021 Datasets and Benchmarks track and receives a Datasets and Benchmarks Award at NeurIPS 2021.

  3. 2023

    FlexVDW for flexible docking

    In 2023, she is first author of FlexVDW, written with Joseph Paggi and Ron Dror and published at the MLDD workshop at ICLR 2023. The deep learning model predicts van der Waals energy while accounting for receptor flexibility, and it improves ligand pose prediction when the protein deforms substantially.

  4. 2024

    A Stanford computer science PhD

    She completes a PhD in Stanford's Computer Science Department in 2024, with Ron Dror as her degree supervisor. Her dissertation, "Efficient learning from 3D molecular structures using equivariant neural networks", covers protein model quality assessment, a machine-learned scoring function for protein-ligand docking with protein flexibility, and pocket-aware 3D fragment-based ligand optimization.

  5. November 2025

    Electron density for 3D learning

    In November 2025, she is first author of "Beyond Atoms: Evaluating Electron Density Representation for 3D Molecular Learning", written at Prescient Design, Genentech with Joshua Rackers, Ewa Nowara, Pedro Pinheiro and colleagues. The paper compares atom-type, raw electron-density and density-gradient voxel inputs for 3D convolutional networks on protein–ligand binding affinity and quantum property prediction.

  6. April 2026

    Language models for drug design

    In April 2026, she coauthors "Evaluating the Progression of Large Language Model Capabilities for Small-Molecule Drug Design" with Shriram Chennakesavalu, Kangway Chuang and colleagues. The paper builds chemistry tasks as reinforcement learning environments and shows that RL post-training lets a smaller model compete with frontier models.

  7. As of October 2026

    Prescient Design at Genentech

    She works at Prescient Design (AI for Drug Discovery), Genentech, in South San Francisco. In July 2026, she coauthors SynLaD, a latent diffusion model that generates synthesizable small molecules conditioned on 3D pharmacophore profiles, which appears in the proceedings of the 43rd International Conference on Machine Learning in 2026.

Career

  1. NowRoche

Education

  1. 2024Stanford UniversityPhD

From the biography.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Roche. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  2. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Roche. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Stanford University. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Stanford University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Patricia Suriana yet. Send us a free-licensed photo.

  1. Learning from Protein Structure with Geometric Vector Perceptrons arxiv.org. Checked 22 October 2026.
  2. ATOM3D: Tasks On Molecules in Three Dimensions arxiv.org. Checked 22 October 2026.
  3. NeurIPS 2021 Awards neurips.cc. Checked 22 October 2026.
  4. FlexVDW: A machine learning approach to account for protein flexibility in ligand docking arxiv.org. Checked 22 October 2026.
  5. Efficient learning from 3D molecular structures using equivariant neural networks | Stanford Digital Repository purl.stanford.edu. Checked 22 October 2026.
  6. Beyond Atoms: Evaluating Electron Density Representation for 3D Molecular Learning arxiv.org. Checked 22 October 2026.
  7. Evaluating the Progression of Large Language Model Capabilities for Small-Molecule Drug Design arxiv.org. Checked 22 October 2026.
  8. SynLaD: Latent Diffusion for Generating Synthesizable Molecules Conditioned on 3D Pharmacophore Profiles arxiv.org. Checked 22 October 2026.
  9. SynLaD: Latent Diffusion for Generating Synthesizable Molecules Conditioned on 3D Pharmacophore Profiles proceedings.mlr.press. Checked 22 October 2026.