Paul Scherer
Machine Learning Scientist at Relation
Greater Cambridge Area United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolPaul Scherer is a Machine Learning Scientist with 11 years of experience bridging academic research and applied graph/temporal ML in industry. He holds a PhD in AI and Computer Science from the University of Cambridge and has translated that expertise into roles from postdoctoral research on multi-omic integration to production-focused ML at Relation. Paul is a hands-on contributor to open-source ML tooling—notably implementing DCRNN and STGCN components for the widely used pytorch_geometric_temporal library—reflecting deep competence in spatiotemporal and graph neural networks. His background spans federated biomedical analysis, active learning libraries, and supervising student research, giving him fluency across experimental design, reproducible software, and deployment-ready models. Colleagues describe him as the kind of researcher-engineer who reliably turns cutting-edge theory into practical, well-tested implementations.
11 years of coding experience
7 years of employment as a software developer
Bachelor of Science (B.Sc.) Joint Artificial Intelligence and Computer Science, Bachelor of Science (B.Sc.) Joint Artificial Intelligence and Computer Science at The University of EdinburghDoctor of Philosophy - PhD Artificial Intelligence and Computer Science, Doctor of Philosophy - PhD Artificial Intelligence and Computer Science at University of Cambridge
English, German, Japanese, Spanish
Github Skills (8)
graph10
pytorch10
deep-learning10
graph-neural-network10
gnn10
python9
tensorflow5
data-engineering4
Programming languages (6)
Github contributions (5)
benedekrozemberczki/pytorch_geometric_temporal
Oct 2020 - May 2021
PyTorch Geometric Temporal: Spatiotemporal Signal Processing with Neural Machine Learning Models (CIKM 2021)
Role in this project:
ML Engineer
ML EngineerContributions:44 commits, 3 PRs, 28 pushes in 6 months
Contributions summary:Paul made significant contributions to the `pytorch_geometric_temporal` repository, which focuses on spatiotemporal signal processing with neural machine learning models. Their work centered on implementing the DCRNN (Diffusion Convolutional Recurrent Neural Network) layer and related components, including its integration, testing, and example usage. They also added STGCN (Spatio-Temporal Graph Convolutional Networks) modules, which included temporal and STConv layers, and created a METR-LA dataset loader, which included data download and preprocessing.
node-embeddingsignalsignal-processingspatio-temporal-analysisgraph-convolution
paulmorio/protclus
Sep 2020 - Nov 2020
Python 3 library implementing a number of topological clustering techniques used on protein-protein interaction networks.
Contributions:53 commits, 1 PR, 7 pushes in 2 months
techniquesinteractionpythonproteinsdocking
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.