About me
I recently earned my Ph.D in Scientific Computing at Uppsala University, and will be joining Professor Samuel Kaski’s Probabilistic Machine Learning group at Aalto University as a postdoctoral researcher. Prior to starting my Ph.D study, I earned MS.c in Computational Science from Uppsala University, MS.c in Chemometrics and BS.c in Chemistry from University of Science and Technology of China.
Research Interests
I am interested in probabilistic methods that operate in latent space while respecting its geometry. Learned representations live on curved and constrained manifolds, and I develop probabilistic and generative methods, largely flow-based, that model distributions on those spaces. Uncertainty quantification for pre-trained vision-language models is where most of this work has landed so far, but it is one application rather than the destination: I am also interested in out-of-distribution generalization and decision making under uncertainty. Earlier in my Ph.D I worked on data heterogeneity problem of federated learning, where my interest in robustness under distributional mismatch began.
News
08/2026: Our paper “PROVE: Probabilistic Visual Grounding via Patch-Level Evidence for Vision-Language Models” got accepted to WACV 2027.
06/2026: I successfully defended my Ph.D thesis, Robust Learning from Distributed and Heterogeneous Data.
05/2026: Our paper “Epistemic Uncertainty Quantification for Pre-trained VLMs via Riemannian Flow Matching” got accepted to ICML 2026.
01/2026: I started an internship at Modulai as a machine learning engineer, studying generalization behavior of reinforcement learning with verifiable rewards for large language models post-training.
09/2025: Our paper “Exploiting the Asymmetric Uncertainty Structure of Pre-trained VLMs on the Unit Hypersphere” got accepted to NeurIPS 2025.
08/2025: Our team Cat Quartet won 2nd place out of >500 teams at Huawei Wireless Communication Global Hackathon 2025 (1st place in Europe region), building a denoising neural SVD operator to approximate SVD operations in a scalable and robust manner.
11/2024: Our team Hello Kitty secured 2nd place out of 35 teams at Huawei Sweden Hackathon 2024, tackling wireless localisation problems using machine learning methods.
06/2024: Our paper “Accelerating Fair Federated Learning: Adaptive Federated Adam” got accepted in IEEE Transactions on Machine Learning in Communications and Networking.
04/2024: Our paper “Federated Learning for Predicting Compound Mechanism of Action Based on Image-data from Cell Painting” got accepted in Artificial Intelligence in the Life Sciences.
01/2024: Our paper “Blades: A Unified Benchmark Suite for Byzantine Attacks and Defenses in Federated Learning” got accepted to IoTDI ‘24.
06/2022: We released Blades, a simulator for Byzantine-robust federated learning with attacks and defenses.
12/21: Our paper “Proactive autoscaling for edge computing systems with kubernetes” got accepted to UCC ‘21.
09/2021: I started my Ph.D study at TDB, Uppsala University, co-supervised by Andreas Hellander, Prashant Singh, and Salman Toor.