Kianoosh Ashouritaklimi
I’m a second year DPhil student in the Department of Statistics at the University of Oxford, supervised by Tom Rainforth and François Caron. Broadly speaking, my research aims to develop more robust and effective sequential decision-making methods for real-world applications. I am also interested in improving the robustness of frequentist UQ methods (e.g. conformal prediction) under common challenges such as model misspecification and distributional shifts.
Before starting my PhD, I completed my MSc in Artificial Intelligence at Imperial College London where I was awarded a Google DeepMind Scholarship. For my dissertation, I analyzed the quality of Bayesian neural network approximations under the supervision of Mark van der Wilk.
I am also passionate about social mobility and widening participation in higher education. Previously, I was an EDI fellow at the University of Oxford, where I had the opportunity to work on the university’s postgraduate recruitment campaigns and initiatives. In my spare time, I also a lousy musician and photographer.
Feel free to reach out to me via email or connect with me on LinkedIn!
News
| Jun 01, 2026 | Check out our new work “Masked Language Flow Models” here. |
|---|---|
| Jun 01, 2026 | Our preprint “Variance-Tilted Diffusion Models for Diverse Sampling” was accepted at the Structured Probabilistic Inference and Generative Modeling Workshop at ICML 2026! |
| Jun 01, 2026 | Our preprint “Safe Particle Selection for Diffusion Steering” was accepted at the Hypothesis Testing Workshop at ICML 2026! |
| Jun 01, 2026 | Our preprint “Prediction-Powered Active Testing” was accepted at the Statistical Frameworks for Uncertainty in Agentic Systems Workshop at ICML 2026! Full version coming out on arXiV soon. |
| May 01, 2026 | Our preprint “Robust Bayes-Assisted Conformal Prediction” was accepted at ICML 2026! |
| Oct 28, 2025 | Our preprint “Active Learning with Task-Driven Representations for Messy Pools” is now available on arXiv. |
| Jan 30, 2025 | Our paper “SymDiff: Equivariant Diffusion via Stochastic Symmetrisation” was accepted at ICLR 2025. |
| Sep 20, 2023 | I started my DPhil at the University of Oxford. |