PeopleofSuperintelligence

Researcher

Jan-Peter Calliess

Jan-Peter Calliess is a machine learning researcher whose work spans control, decision-making and finance. He coauthors research on multi-robot collision avoidance, earns a DPhil in Oxford Engineering Science, and serves as a research associate in Cambridge. His later papers apply deep reinforcement learning and autoencoders to trading, and in 2026 he coauthors Canary, a method for value-at-risk constrained reinforcement learning. As of October 2026, he is a Senior Research Fellow in Oxford's Department of Engineering Science.

Works
47
Citations
745
h-index
14
Badges

Publication numbers from OpenAlex, as of 27 October 2026.

Timeline

  1. 2011

    Multi-robot collision avoidance

    In 2011, he coauthors the paper “Lazy auctions for multi-robot collision avoidance and motion control under uncertainty.”

  2. 2014

    A DPhil in Engineering Science

    His undergraduate degrees are from the University of Karlsruhe, and he serves as a research associate in Cambridge’s Control Group and Computational and Biological Learning Lab. In 2014, he completes a DPhil in Engineering Science at Oxford on conservative decision-making and inference in uncertain dynamical systems.

  3. 2017

    Lipschitz interpolation for learning

    His research interests include machine learning, decision-making with applications to finance, and control. In 2017, he proposes optimizing the assumed input metric for Lipschitz interpolation by minimizing validation-set prediction error.

  4. 2023

    Deep Q-learning for order execution

    In 2023, he coauthors a deep reinforcement-learning agent that translates a high-frequency trading signal into a strategy for placing individual limit orders, using an ABIDES-based NASDAQ simulator.

  5. 2024

    Autoencoder for statistical arbitrage

    In 2024, he and Fabian Krause train a standard autoencoder on US stock returns to derive trading strategies based on the Ornstein–Uhlenbeck process.

  6. 2026

    Canary for safe reinforcement learning

    In 2026, he and Rohan Tangri introduce Canary, a method that uses Cantelli’s inequality to create a smooth, conservative bound on value-at-risk constraints in reinforcement learning.

  7. As of October 2026

    Senior Research Fellow at Oxford

    As of October 2026, he is a Senior Research Fellow in the Department of Engineering Science at the University of Oxford.

Career

  1. NowUniversity of OxfordSenior Research Fellow
  2. University of CambridgeResearch Associate in the Control Group and Computational and Biological Learning Lab

Education

  1. 2014University of OxfordDPhil in Engineering Science
  2. Karlsruhe Institute of TechnologyUndergraduate degrees

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Jan-Peter Calliess yet. Send us a free-licensed photo.

  1. Jan-Peter Calliess robots.ox.ac.uk. Checked 27 October 2026.
  2. Conservative decision-making and inference in uncertain dynamical systems - ORA - Oxford University Research Archive ora.ox.ac.uk. Checked 27 October 2026.
  3. Jan-Peter Calliess – MLG Cambridge mlg.eng.cam.ac.uk. Checked 27 October 2026.
  4. Asynchronous Deep Double Duelling Q-Learning for Trading-Signal Execution in Limit Order Book Markets arxiv.org. Checked 27 October 2026.
  5. End-to-End Policy Learning of a Statistical Arbitrage Autoencoder Architecture arxiv.org. Checked 27 October 2026.
  6. Cantelli Constrained Policy Optimization arxiv.org. Checked 27 October 2026.
  7. Jan-Peter Calliess | ELLIS Oxford ellis.ox.ac.uk. Checked 27 October 2026.