Benedek Rozemberczki
Staff Research Scientist at Google
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolBenedek Rozemberczki is a Staff Research Scientist with a decade of experience building and shipping graph machine learning systems, currently at Google after a research-focused tenure at Isomorphic Labs. He holds a PhD in Data Science from the University of Edinburgh and combines deep academic grounding with hands‑on PyTorch engineering—authoring and contributing to influential open-source graph libraries such as PyTorch Geometric and Karate Club. His work spans graph representation learning, spatio-temporal GNNs, and scalable training algorithms (Cluster-GCN, SimGNN, graph2vec), reflecting both model-building and dataset engineering expertise. Comfortable moving ideas from papers to production, he has implemented canonical benchmarks and novel datasets (e.g., GitHub, Twitch, Pathfinder) that broaden real-world GNN evaluation. Based in London, he brings a rare mix of research rigor, practical coding impact, and domain breadth informed by economics and data science training.
10 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD, Data Science, Doctor of Philosophy - PhD, Data Science at The University of EdinburghMaster of Arts (M.A.), Economic Policy, Master of Arts (M.A.), Economic Policy at Central European UniversityBachelor of Arts (B.A.), Applied Economics, Bachelor of Arts (B.A.), Applied Economics at Corvinus University of Budapest
English, Hungarian
Github Skills (21)
geometric-deep-learning10
pytorch10
lib10
python10
gnn10
clustering-algorithm10
data-set10
attention-mechanism10
machine-learning10
networkx10
datasets10
deeplearning-ai10
graph-embedding10
deep-learning10
graph-neural-network10
Programming languages (14)
Github contributions (5)
benedekrozemberczki/graph2vec
Jul 2018 - Nov 2022
A parallel implementation of "graph2vec: Learning Distributed Representations of Graphs" (MLGWorkshop 2017).
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 85 commits, 3 PRs in 4 years 4 months
Contributions summary:Benedek primarily contributed to the core functionality of the graph2vec project. Their commits involved modifying the `graph2vec.py` and `parser.py` files, which indicate a focus on feature extraction, model training, and parameter parsing for the graph embedding algorithm. The changes included implementing the Weisfeiler-Lehman feature extraction, updating the parameter parser, and saving the embedding, suggesting an active role in refining the model's implementation and its operational aspects.
kernelnode-embeddingdistributed-representationsword2vecstruc2vec
benedekrozemberczki/CapsGNN
Jan 2019 - Nov 2022
A PyTorch implementation of "Capsule Graph Neural Network" (ICLR 2019).
Role in this project:ML Engineer
ML EngineerContributions:1 release, 177 commits, 3 PRs in 3 years 9 months
Contributions summary:Benedek contributed significantly to the implementation and refinement of a Capsule Graph Neural Network (CapsGNN) model. Their commits encompass core model architecture definition, including the integration of GCN layers, primary and secondary capsule layers, and an attention mechanism. The user was also involved in defining and integrating the margin loss function, crucial for training the capsule networks, as well as integrating the histogram reconstruction layers and their corresponding loss calculations.
graph-neural-networkpytorchcapsgnncapsule-networkcapsule-neural-networks
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.