Ondrej Bohdal

I'm a senior machine learning researcher at Samsung Research, where I focus on personalization of generative AI models. Before joining Samsung, I was a postdoctoral researcher at the University of Edinburgh, working on topics such as multimodal large language models, diffusion models, fairness, uncertainty calibration and out-of-distribution generalization.

I did my PhD on Meta-Learning Algorithms and Applications at the University of Edinburgh, advised by Timothy Hospedales. I was a research intern at Samsung AI Center, Cambridge and Amazon Web Services, Berlin, and also did part of my studies at the Alan Turing Institute in London.

Email  /  Google Scholar  /  Twitter  /  Github  /  LinkedIn  /  London, UK

Research Overview / Publications / Background

profile photo

Publications
CG-TTRL: Context-Guided Test-Time Reinforcement Learning for On-Device Large Language Models
Peyman Hosseini, Ondrej Bohdal, Taha Ceritli, Ignacio Castro, Matthew Purver, Mete Ozay, Umberto Michieli
Under review, 2025
paper
K-Merge: Online Continual Merging of Adapters for On-device Large Language Models
Donald Shenaj, Ondrej Bohdal, Taha Ceritli, Mete Ozay, Pietro Zanuttigh, Umberto Michieli
Under review, 2025
paper
Efficient Compositional Multi-tasking for On-device Large Language Models
Ondrej Bohdal, Mete Ozay, Jijoong Moon, Kyeng-Hun Lee, Hyeonmok Ko, Umberto Michieli
EMNLP, 2025
paper / project page / code / blog
On-device System of Compositional Multi-tasking in Large Language Models
Ondrej Bohdal, Konstantinos Theodosiadis, Asterios Mpatziakas, Dimitris Filippidis, Iro Spyrou, Christos Zonios, Anastasios Drosou, Dimosthenis Ioannidis, Kyeng-Hun Lee, Jijoong Moon, Hyeonmok Ko, Mete Ozay, Umberto Michieli
EMNLP (Industry Track), 2025
paper / demo
HydraOpt: Navigating the Efficiency-Performance Trade-off of Adapter Merging
Taha Ceritli, Ondrej Bohdal, Mete Ozay, Jijoong Moon, Kyeng-Hun Lee, Hyeonmok Ko, Umberto Michieli
EMNLP, 2025
paper
LoRA.rar: Learning to Merge LoRAs via Hypernetworks for Subject-Style Conditioned Image Generation
Donald Shenaj, Ondrej Bohdal, Mete Ozay, Pietro Zanuttigh, Umberto Michieli
ICCV, 2025
Short version won the best paper award at the ICCV 2025 Personalization in Generative AI Workshop!
paper / project page / code / blog
VL-ICL Bench: The Devil in the Details of Benchmarking Multimodal In-Context Learning
Yongshuo Zong*, Ondrej Bohdal*, Timothy Hospedales
* Joint first authors
ICLR, 2025
paper / project page / code / data
MemControl: Mitigating Memorization in Medical Diffusion Models via Automated Parameter Selection
Raman Dutt, Ondrej Bohdal, Pedro Sanchez, Sotirios A. Tsaftaris, Timothy Hospedales
WACV, 2025
paper
On the Limitations of General Purpose Domain Generalisation Methods
Henry Gouk, Ondrej Bohdal, Da Li, Timothy Hospedales
Under review, 2024
paper
Memorized Images in Diffusion Models share a Subspace that can be Located and Deleted
Ruchika Chavhan, Ondrej Bohdal, Yongshuo Zong, Da Li, Timothy Hospedales
ICML GenLaw workshop, 2024
paper
Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models
Yongshuo Zong, Ondrej Bohdal, Tingyang Yu, Yongxin Yang, Timothy Hospedales
ICML, 2024
paper / project page / code / data
Navigating Noise: A Study of How Noise Influences Generalisation and Calibration of Neural Networks
Martin Ferianc*, Ondrej Bohdal*, Timothy Hospedales, Miguel Rodrigues           
* Joint first authors
TMLR, 2024
paper / code / video
FairTune: Optimizing Parameter Efficient Fine Tuning for Fairness in Medical Image Analysis
Raman Dutt, Ondrej Bohdal, Sotirios A. Tsaftaris, Timothy Hospedales
ICLR, 2024
paper / code
Feed-Forward Latent Domain Adaptation
Ondrej Bohdal, Da Li, Shell Xu Hu, Timothy Hospedales
WACV, 2024
paper / project page / video / slides / patent
Meta-Calibration: Learning of Model Calibration Using Differentiable Expected Calibration Error
Ondrej Bohdal, Yongxin Yang, Timothy Hospedales
TMLR, 2023
paper / code
Meta Omnium: A Benchmark for General-Purpose Learning-to-Learn
Ondrej Bohdal*, Yinbing Tian*, Yongshuo Zong, Ruchika Chavhan, Da Li, Henry Gouk, Li Guo, Timothy Hospedales
* Joint first authors
CVPR, 2023
paper / project page / code / video / slides / poster
PASHA: Efficient HPO and NAS with Progressive Resource Allocation
Ondrej Bohdal, Lukas Balles, Martin Wistuba, Beyza Ermis, Cédric Archambeau, Giovanni Zappella
ICLR, 2023
HPO: hyperparameter optimization, NAS: neural architecture search
paper / code / tutorial / video / slides / poster
Label Calibration for Semantic Segmentation Under Domain Shift
Ondrej Bohdal, Da Li, Timothy Hospedales
ICLR Trustworthy ML workshop, 2023
paper
Fairness in AI and Its Long-Term Implications on Society
Ondrej Bohdal, Timothy Hospedales, Philip H.S. Torr, Fazl Barez
Stanford Existential Risks Conference, 2023
paper
Feed-Forward Source-Free Domain Adaptation via Class Prototypes
Ondrej Bohdal, Da Li, Timothy Hospedales
ECCV OOD-CV workshop, 2022
paper
EvoGrad: Efficient Gradient‑Based Meta‑Learning and Hyperparameter Optimization
Ondrej Bohdal, Yongxin Yang, Timothy Hospedales
NeurIPS, 2021
paper / blog / code / video / slides / poster
A Channel Coding Benchmark for Meta‑Learning
Rui Li, Ondrej Bohdal, Rajesh Mishra, Hyeji Kim, Da Li, Nicholas Lane, Timothy Hospedales
NeurIPS (datasets and benchmarks track), 2021
paper / code / blog / video
Flexible Dataset Distillation: Learn Labels Instead of Images
Ondrej Bohdal, Yongxin Yang, Timothy Hospedales
NeurIPS MetaLearn workshop, 2020
paper / code / video
Semantic Segmentation of 3D Point Clouds
Data study group at the Alan Turing Institute, 2020
report

Design and source code from Jon Barron's website