Thomas Tumiel
Research Engineer (Tech Lead) at Google DeepMind
United Kingdom
Summary
👤
Senior🎓
Top SchoolThomas Tumiel is a Research Engineer and Tech Lead at Google DeepMind with 10 years' experience translating frontier ML research into real-world impact, currently leading post-training work on large language models across reasoning, creativity, and scientific discovery. He has hands-on expertise in reasoning agents, world models, combinatorial optimisation and interpretability, and previously built reinforcement-learning systems for production PCB design at InstaDeep. Thomas combines deep research instincts with production engineering—evidenced by contributions like refining operation counting for PyTorch models—to ensure reproducible, deployable ML systems. He cut his teeth building end-to-end projects from a NAND-based computer to mobile apps and led an AI/crypto student society of ~300 members, showing a blend of curiosity, teaching, and community leadership.
10 years of coding experience
4 years of employment as a software developer
Bachelor of Engineering (BEng), Electrical and Electronics Engineering, Bachelor of Engineering (BEng), Electrical and Electronics Engineering at University of Cape Town
Github Skills (11)
pytorch10
machine-learning10
deep-learning10
optimization9
batch-normalization9
optimisation9
numerical-optimization9
algorithm9
code-optimization9
algorithms9
python8
Programming languages (9)
Github contributions (5)
Lyken17/pytorch-OpCounter
Mar 2019 - Mar 2019
Count the MACs / FLOPs of your PyTorch model.
Role in this project:
ML Engineer
ML EngineerContributions:6 commits, 1 PR in 1 day
Contributions summary:Thomas primarily focused on adding and refining functionality related to operations counting for PyTorch models, specifically for adaptive pooling and batch normalization. Their contributions included implementing the counting logic for adaptive average and max pooling layers, and ensuring the accurate calculation of operations. They also addressed potential issues related to singleton sizes and fixed the total operations calculations to improve repeatability in runs.
pytorch
ttumiel/Crypto-Resources
Sep 2017 - Feb 2018
Contributions:14 commits, 1 PR, 12 pushes in 4 months
deckexpertblockchaincryptocurrencyblockchain-resources
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.