I’m excited about the potential for AI to accelerate the way we do science. We started Diffractive Labs to do just that.

I just defended my PhD, where I looked at how AI can accelerate discovery cycles across domains and parts of the scientific method: from experimental design to simulation and reasoning.

Studied at: Oxford (PhD ML, with Yarin Gal & Steve Roberts @ OATML) · UCL (MSc ML) · Cambridge (MSci Physics)

Worked at: Normal Computing · NASA FDL · Humanloop (acq. Anthropic) · Caltech (Physics)

Publications

  1. The Role of Absorption in Three-Dimensional Electron Diffraction Dynamical Structure Refinement

    B. Colmey, T. A. S. Doherty, S. A. Malik, P. A. Midgley · 2026 · arXiv preprint (under review)

    Paper

  2. MADE: Benchmark Environments for Closed-Loop Materials Discovery

    S. A. Malik, T. Doherty, P. Tigas, M. Razzak, S. J. Roberts, A. Walsh, Y. Gal · 2026 · ICML

    PaperCode

  3. Hybrid Physics-Machine Learning Models for Quantitative Electron Diffraction Refinements

    S. A. Malik*, T. A. S. Doherty*, B. Colmey, S. J. Roberts, Y. Gal, P. A. Midgley · 2026 · Nature Communications

    PaperCode

  4. Accelerating Long-period Exoplanet Discovery by Combining Deep Learning and Citizen Science

    S. A. Malik, N. L. Eisner, I. R. Mason, S. Platymesi, S. Aigrain, S. J. Roberts, Y. Gal, C. J. Lintott · 2025 · The Astronomical Journal

    PaperCode

  5. Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs

    J. Kossen, J. Han, M. Razzak, L. Schut, S. A. Malik, Y. Gal · 2024 · arXiv preprint

    Paper

  6. High-Cadence Thermospheric Density Estimation Enabled by Machine Learning on Solar Imagery

    S. A. Malik*, J. Walsh*, G. Acciarini, T. E. Berger, A. G. Baydin · 2023 · NeurIPS ML4PS Workshop

    Paper

  7. BatchGFN: Generative Flow Networks for Batch Active Learning

    S. A. Malik, S. Lahlou, A. Jesson, M. Jain, N. Malkin, T. Deleu, Y. Bengio, Y. Gal · 2023 · ICML SPIGM Workshop

    PaperCode

  8. Discovering Long-period Exoplanets Using Deep Learning with Citizen Science

    S. A. Malik, N. L. Eisner, C. J. Lintott, Y. Gal · 2022 · NeurIPS ML4PS Workshop

    PaperCode

  9. Multi-Modal Fusion by Meta-Initialization

    M. T. Jackson*, S. A. Malik*, M. T. Matthews, Y. Mohamed-Ahmed · 2022 · FARSCOPE Robotics Workshop · Best Paper Award

    PaperCode

  10. Predicting the Outcomes of Material Syntheses with Deep Learning

    S. A. Malik, R. E. A. Goodall, A. A. Lee · 2021 · Chemistry of Materials

    PaperCode