– Present
Postdoctoral Researcher
· Eindhoven University of Technology
Efficient adaptation and inference for FMs, and evaluation of LLM reasoning.
Working with M. Holenderski.
Academic Curriculum Vitae
Vasileios Tsouvalas
Postdoctoral ResearcherEindhoven University of Technology (TU/e) · Netherlands
Research Interests
- Efficient multimodal foundation models on edge
- Reliable & decentralized learning · FL and SL
- Privacy, security and fairness in edge ML
- Evaluation of LLM reasoning
Experience & Education
–
PhD
· Eindhoven University of Technology
Data and communication efficiency in federated learning.
Supervisors: N. Meratnia and T. Özçelebi.
Thesis: Chaos to Harmony: Taming Heterogeneity and Communication Bottlenecks in Federated Learning.
–
MSc in Embedded Systems
· Eindhoven University of Technology
Thesis Distinction Research on semi-supervised federated learning for audio recognition in IRIS.
Supervisors: T. Özçelebi and A. Saeed.
–
Diploma (Integrated MSc) in Electrical and Computer Engineering
AUTh · Aristotle University of Thessaloniki
Thesis Distinction · 10/10 Research on joint autonomous exploration with aerial and ground robots at ARL. Thesis (Greek).
Joint thesis with K. Baxevani. Thesis supervisors: E. Tsardoulias and L. Petrou.
Teaching & Supervision
Teaching & Practical Instruction
Student Research Supervision
-
SDAPT · Domain-adaptive pretraining of vision foundation models through split learning.
-
AE-PSL · Autoencoder-compressed parallel split learning for model fine-tuning.
-
MPSL · Fine-tuning multimodal transformers through parallel split learning.
-
Matthijs MarinusUbiSec 2024
FiBA · Invisible backdoor attacks in federated learning.
-
DA-MLDG · Meta-learning and domain generalization for speech emotion recognition.
-
Federated-learning communication on the OKdo E1 microcontroller.
Research Talks
EncCluster: Scalable Functional Encryption in Federated Learning
Research Talk · Joint Work
With S. Mohammadi (RISE).
Labeling Chaos to Learning Harmony: Federated Learning with Label Noise
Research Talk
Conferences & Workshops
ECML-PKDD 2026
WAFL Workshop
AE-PSL: autoencoder-compressed parallel SL for FM fine-tuning.
CIKM 2025
EFU: federated unlearning through functional encryption.
IJCAI 2025
2 Papers FedGenAI Workshop
MaTU: many-task federated fine-tuning with unified task vectors; MPSL: multimodal FM adaptation through parallel SL.
UbiSec 2024
FiBA: invisible backdoor attacks in FL.
NeurIPS 2024
FL@FM Workshop
EncCluster: scalable functional encryption for FL with FMs.
IEEE Big Data 2024
DeltaMask: communication-efficient federated adaptation of VFMs.
ICML 2024
FM-Wild Workshop
DeltaMask: communication-efficient federated adaptation of VFMs.
IEEE EDGE 2024
FedCode: communication-efficient FL through codebook transfer.
ICASSP 2024
FedCompress: communication-efficient FL through weight clustering and server-side distillation.
PerCom 2023
PerFail Workshop
DA-MLDG: meta-learning for domain generalization in speech emotion recognition.
NeurIPS 2022
INTERPOLATE Workshop
FedLN: robust FL with noisy client labels.
ICASSP 2022
FedSTAR: semi-supervised FL for audio recognition.
PerCom 2022
EmotionAware Workshop
FedSER: semi-supervised FL for speech emotion recognition.
Academic Service
- Technical Committee Member
(ML for Signal Processing)
- AgenticOS @ NeurIPS 2026WorkshopTechnical Committee Member
- ACL ARR 2026August 2026 CycleExternal Reviewer
- ECML-PKDD 2026Research TrackExternal Reviewer
- CFAgentic @ ICML 2025WorkshopTechnical Committee Member
- FLute @ ICASSP 2025Federated Learning for Audio Understanding · WorkshopTechnical Committee Member
Awards & Distinctions
Best Master's Thesis Award
Department of Mathematics and Computer Science, TU/e.
Diploma Thesis Distinction 10/10
Electrical and Computer Engineering, Aristotle University of Thessaloniki (AUTh).
Aerial scouting and ground navigation for autonomous exploration in RoboCup Rescue environments.
Top 1% National STEM Exams
Greek Panhellenic university entrance examinations · 18,800/20,000.
Professional Development
Model Parallelism · Building & Deploying Large Neural Networks
NVIDIA Deep Learning Institute · Certificate of Competency
Training and deployment across multiple GPUs. Instructor: H. Shoukourian.
Eindhoven, The Netherlands
Data Parallelism · Training Deep Learning Models on Multiple GPUs
NVIDIA Deep Learning Institute · Certificate of Competency
Distributed training with PyTorch DistributedDataParallel (DDP). Instructor: H. Shoukourian.
Eindhoven, The Netherlands
ASCI · Doctoral Education Certificate
Advanced School for Computing and Imaging
Completed the ASCI doctoral education programme alongside my PhD thesis, Chaos to Harmony.
ASCI A25 · Computer Vision by Learning
Advanced School for Computing and Imaging
Vision foundation models (VFMs), self-supervised learning, 3D vision and video understanding.
Course lead: C. Snoek. Invited tutorial: Y. Asano.
Amsterdam Science Park, Netherlands
DeepLearn 2024 · 11th International School on Deep Learning
Programme highlights:
- L. Benini · Open hardware for edge machine learning
- G. Karypis · LLM inference optimization
- F. Hutter · Automated machine learning (AutoML)
Porto–Maia, Portugal
ICT with Industry 2024 · LLMs & Model-Based Testing
Lorentz Center
LLM-assisted model testing for Axini at the Lorentz Center. Academic lead: D. Sapra.
Leiden, Netherlands
–
ASCI A16 · Winter School on Efficient Deep Learning
Advanced School for Computing and Imaging
Energy-efficient hardware, spiking neural networks, and distributed and federated learning.
Course leads: H. Corporaal, L. Chen and A. Pimentel.
Leiden, Netherlands
Publications
Papers, code and project pages, searchable by topic, method or author.
Explore Publications