Tim Dettmers

Ticino, Switzerland

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Tim Dettmers is a machine learning engineer with 12 years of experience who focuses on practical, deployable AI—particularly efficient large-model finetuning and knowledge-graph embeddings. Moving from a non-traditional educational start to degrees in mathematics, statistics and computer science, he blends rigorous theory with hands-on systems work, contributing to notable projects like ConvE and the bitsandbytes quantization tooling. His contributions span model implementation and evaluation (ConvE, DistMult, Complex), improving training pipelines (QLoRA), and broadening deployment support across CUDA versions for bitsandbytes, reflecting both ML expertise and DevOps fluency. Persistent, collaborative, and privacy-minded, he thrives in quiet, cooperative environments where his work directly benefits others.

12 years of coding experience

Bachelor of Science (Honours), Mathematics and Statistics, --, Bachelor of Science (Honours), Mathematics and Statistics, -- at The Open University

Master’s Degree, Computer Science, Master’s Degree, Computer Science at Università della Svizzera italiana

Mathematical-technical Software Developer, Computer Science, Mathematical-technical Software Developer, Computer Science at Oberstufenzentrum für Informations- und Medizintechnik Berlin-Neukölln

Github Skills (20)

transformers10

pytorch10

convolutional-neural-networks10

python10

scripting10

machine-learning10

cicd10

script10

fine-tuning10

neural-network10

lora10

cuda10

sh10

shell10

devops10

Programming languages (3)

C++Jupyter NotebookPython

Github contributions (5)

bitsandbytes-foundation/bitsandbytes

Jun 2021 - Jan 2023

Accessible large language models via k-bit quantization for PyTorch.

Role in this project:

DevOps Engineer

Contributions:10 releases, 25 reviews, 390 commits in 1 year 7 months

Contributions summary:Tim's commits focused on modifying and improving the deployment scripts for the `bitsandbytes` repository. They updated the `deploy_from_slurm.sh` script to support CUDA versions ranging from 9.2 to 12.2, including both standard and no-matmul builds. The changes involved the building and deploying of precompiled binaries. These changes indicate a focus on improving the software distribution process to ensure support for a wide range of CUDA versions.

cudapytorch8-bitgpupruning

TimDettmers/ConvE

Jun 2017 - Oct 2022

Convolutional 2D Knowledge Graph Embeddings resources

Role in this project:

ML Engineer

Contributions:63 commits, 4 PRs, 46 pushes in 5 years 4 months

Contributions summary:Tim primarily contributed to a knowledge graph embedding project, implementing and modifying various models for link prediction. Their work included adding datasets, preprocessing scripts, and model implementations (ConvE, DistMult, and Complex). They also focused on evaluating the models, including implementing a reverse rule evaluation, demonstrating a strong understanding of the project's core machine learning tasks.

knowledge-graph-embeddingsknowledgeembeddingsconvolutionalgraph-embeddings

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.

Request Free Trial