Theodore Popp
Senior Member Of Technical Staff at AMD
Munich, Bavaria, Germany
Summary
🤩
Rockstar🎓
Top SchoolTheodore Popp is a compiler-focused software engineer with 11 years of experience building and refining ML-centric compiler toolchains, currently a Senior Member of Technical Staff at AMD based in Munich. He has deep hands-on expertise with MLIR/LLVM, XLA, and runtimes—contributing significant refactors and LLVM integrations to high-profile projects like IREE, TensorFlow, and OpenXLA. Theodore’s work spans low-level GPU and accelerator compilers (including early Pixel Visual Core efforts) to runtime compatibility and test infrastructure, demonstrating both systems-level rigor and attention to build/release stability. Colleagues rely on him to untangle API breakages, modernize dependency workflows, and improve numerical correctness across devices—skills reflected in repeated upstream contributions that keep complex ML stacks reproducible and performant.
11 years of coding experience
10 years of employment as a software developer
Distinguished Achievement Texas plan n/a, Distinguished Achievement Texas plan n/a at Tivy High SchoolB.S Computer Science, B.S Computer Science at University of Texas
English, French
Stackoverflow
846reputation
94kreached
4answers
0questions
tensorflow
top-5%
deep-learning
top-5%
keras
top-5%
Github Skills (24)
c-language10
llvm10
machine-learning10
mlr10
deeplearning-ai10
compiler-design10
compiler-compiler10
deep-learning10
tensorflow10
bazel10
xla10
compiler10
cprogramming-language10
python9
neural-network9
Programming languages (6)
Github contributions (5)
tensorflow/runtime
Jun 2020 - Jan 2023
A performant and modular runtime for TensorFlow
Role in this project:
Back-end Developer
Back-end DeveloperContributions:59 commits in 2 years 7 months
Contributions summary:Theodore primarily integrated and updated LLVM versions within the TensorFlow runtime project. Their contributions involved modifying the `dependencies.bzl` and `third_party/llvm/workspace.bzl` files to incorporate specific LLVM commit hashes and SHA256 checksums. This work ensures the TensorFlow runtime project remains compatible with and benefits from the latest LLVM features. The user also added dependent dialects and migrated MLIR instruction member accesses to prefixed form.
runtimeperformantmodulartensorflow
openxla/xla
Jan 2020 - Jan 2023
A machine learning compiler for GPUs, CPUs, and ML accelerators
Role in this project:Back-end Developer
Back-end DeveloperContributions:12 reviews, 53 commits, 13 PRs in 3 years
Contributions summary:Theodore primarily focused on enhancing the XLA compiler, specifically improving test validity and HLO printing to adhere to the JSON specification. They updated missing dependencies to support TensorFlow builds without a GPU. Further contributions included updating the MLIR API and modifying gather clamping optimizations to prevent incorrect type casts, as well as improving the accuracy of device-specific tanh functions for f64 inputs.
compilercommunity-drivenmachine-learningmodular
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.