Curriculum vitae

CV

Education, publications, research and industry experience, and activities.

Education

  1. 2025 – present

    Ph.D. in Computer Science

    Princeton University

  2. 2022 – 2023

    M.Sc. in Advanced Computer Science

    University of Oxford

  3. 2018 – 2021

    B.Sc. in Mathematics & Computer Science

    École Polytechnique, Summa Cum Laude

  4. 2019

    International Honors Program

    Stanford University

Publications

  1. NeurIPS 2026

    SCION: Scene Composition with Instanced Neural Primitives

    William Koch, Amogh Joshi, Cyrus Vachha, Cheng Zheng, Felix Heide

    ProjectarXiv

  2. ECCV 2026

    ScenarioControl: Vision-Language Controllable Vectorized Latent Scenario Generation

    Lili Gao*, Yanbo Xu*, William Koch*, Samuele Ruffino, Luke Rowe, Behdad Chalaki, Dmitriy Rivkin, Julian Ost, Roger Girgis, Mario Bijelic, Felix Heide

    ProjectarXivGitHubBlog post

  3. NeurIPS 2025

    HEIR: Learning Graph-Based Motion Hierarchies

    William Koch*, Cheng Zheng*, Baiang Li, Felix Heide

    ProjectarXivGitHub

  4. 2023

  5. 2020

* Equal contribution

Experience

  1. Aug 2025 – present

    PhD Researcher

    Princeton Computational Imaging Lab, Princeton University

    • Researching world models, in particular controllable video generation with long-horizon memory, advised by Felix Heide.
    • Developing spatially disentangled latent representations, designing and training spatial autoencoders and diffusion models from scratch with multi-GPU training.
    • Led the controllable video-generation branch of ScenarioControl (ECCV 2026) and the graph-based motion-hierarchy research of HEIR (NeurIPS 2025); first author of SCION (NeurIPS 2026).

  2. 2025 – present

    Graduate Fellow

    AI for Accelerating Invention (AI²), Princeton AI Lab

  3. Nov 2023 – Jul 2025

    Associate

    Boston Consulting Group GmbH

  4. 2023

    ML Research Intern

    Active Vision Lab, University of Oxford

    • Research on non-rigid neural radiance fields for virtual-reality video conferencing with the Department of Engineering; supervised by Ismail Ceylan, Victor Prisacariu, and Philip Torr.
    • Developed an efficient novel-pose and view synthesis model that turns standard webcam input into new perspectives of the person in front of the camera.
    • Proposed a lightweight preprocessing stage for training and inference that uses existing priors to optimize ray sampling for real-time performance.

  5. Jan – Jun 2022

    Applied Science Intern

    Zalando Switzerland AG

    • Designed and implemented generative adversarial networks; initiated and ran experiments on unsupervised image inpainting.
    • Contributed to the architecture of 3D generative models trained without 3D data by leveraging suggestive contours.
    • Improved the performance of existing image tooling and extended a shared preprocessing library (semantic segmentation, dense UV unwrapping of humans).
    • Built an application for automatically generating datasets of model images with desired attributes.

  6. 2021

    Consulting Intern

    d-fine GmbH

    • Six-month internship on two projects, sustainable urban transport and decentralized cloud infrastructure using distributed ledger technologies.

  7. 2020

    Cryptography Research Intern

    Télécom Paris

    • Thesis on privacy-preserving face recognition for large-scale video surveillance using secure multi-party computation; proposed a framework that makes mass surveillance cryptographically impossible, with performance limitations.

  8. Jul – Aug 2020

    Cryptography Research Intern

    Deutsche Bank AG

    • Researched machine learning on encrypted data using homomorphic encryption and implemented a proof of concept.

Skills

Programming
Python, C++, CUDA, Rust, C
Machine learning
Transformers, diffusion models, autoencoders, representation learning, neural rendering, graph neural networks
Tools
PyTorch, PyTorch Geometric, PyTorch DDP, Weights & Biases
Systems
Multi-GPU training, CUDA kernel optimization, SLURM, Linux, Docker, Git, tmux
Claude workflows
Claude Code (skills, hooks), Claude Cowork, research automation, cluster-job orchestration
Languages
English (native), German (native), French (B2–C1)

Activities

  1. 2025 – present

    Graduate Fellow

    Princeton AI Lab's AI for Accelerating Invention (AI²) initiative

  2. 2018 – 2023

    Scholar

    Studienstiftung des deutschen Volkes

    • Scholarship spokesperson for the Paris region (2019 – 2020).
    • Organized a one-week academy on state surveillance, filter bubbles, and big data (April 2022) with a budget of nearly $40,000, 80 participants, 6 researchers, and further speakers.
    • Organized and moderated events on educational equality and digital independence.

  3. Jan 2019 – Jun 2021

    Founder and President

    Privacy Club, École Polytechnique

  4. Mar – Jun 2021

    Co-Founder and Secretary

    IEEE Student Branch, École Polytechnique

  5. Dec 2020 – Jun 2021

    Communication

    Robotics Club, École Polytechnique

  6. Software Developer, DESTINY

    REXUS/BEXUS student research program, École Polytechnique

    • Detection of Earthquakes through a STratospheric INfrasound studY; voluntary student research project.

  7. Contributor

    Binet Réseau (IT club), École Polytechnique

    • Projects including a self-hosted XMPP server and Nextcloud instance for students.