Nikolaos Dimitriadis
Doctoral Assistant at EPFL
Lausanne, Vaud, Switzerland
Summary
👤
Senior🎓
Top SchoolNikolaos Dimitriadis is a Machine Learning researcher and Doctoral Assistant at EPFL with nine years of experience and a PhD focused on how knowledge is encoded and transferred in the weight space of foundation models. His work spans Pareto front parametrization for multi-task learning (PaMaL, PaLoRA) and practical model-merging techniques that localize task-specific information and mitigate forgetting (Tall-Masks, LiNeS, MEMOIR). He completed two DeepMind internships tackling visual text rendering in diffusion models and multi-task post-training for large language models, scaling experiments up to 27B parameters. Nikolaos combines rigorous theoretical contributions with scalable, low-rank and post-training methods that improve downstream performance of large pre-trained models, and his research often uncovers compact, task-local structures within merged weights.
9 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at EPFLMaster of Engineering - MEng, Electrical and Computer Engineering, Master of Engineering - MEng, Electrical and Computer Engineering at National Technical University of Athens
Chinese, English, French, Greek
Github Skills (9)
multi-task-learning9
inference3
compression3
java2
machine-learning2
post-training2
gps2
gps-tracking1
artificial-intelligence1
Programming languages (2)
Github contributions (5)
nik-dim/gps
Nov 2017 - Apr 2020
Contributions:2 PRs, 12 pushes, 4 branches in 2 years 5 months
artificialmachine-learningtrackinggpsgps-tracking
nik-dim/pamal
Jun 2023 - Jun 2026
Official repository of "Pareto Manifold Learning: Tackling multiple tasks via ensembles of single-task models" [ICML 2023]
Contributions:2 pushes, 1 branch, 2 comments in 3 years 1 month
icml-2023mtlmulti-task-learningmultitask-learningpareto-front
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.