Gheorghe-teodor Bercea
Principal Member Of Technical Staff at AMD
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolGheorghe-teodor Bercea is a Principal Member of Technical Staff at AMD with 11 years of experience building high-performance compilers and GPU-accelerated software from research prototypes to production-grade implementations. He holds a PhD from Imperial College London and has a strong research-to-industry trajectory through roles at IBM and AMD, bridging deep compiler theory with practical backend engineering. His open-source contributions include key improvements to Clang’s OpenMP/CUDA offloading, ONNX-MLIR lowering in the Krnl dialect, and ROCm-focused PyTorch kernel optimizations, demonstrating expertise in GPU code generation and performance tuning. He has also worked on build and automation tooling for cloud-native integration via Apache Camel K, showing attention to developer ergonomics and deployment workflows. Colleagues rely on him for robust low-level solutions that make advanced parallelism and device-specific optimizations reliable in real-world systems. An understated strength is his ability to translate academic compiler techniques into tangible runtime and kernel improvements that measurably boost performance on AMD and NVIDIA platforms.
11 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Imperial College LondonPOLITEHNICA București National University for Science and Technology
English, Romanian, French
Github Skills (42)
camelcase10
pytorch10
docker10
c-language10
back-end-development10
operation10
tensorrt10
onnx10
mlr10
apache-camel10
dockers10
openmp10
automation10
roc10
go10
Programming languages (9)
Github contributions (5)
onnx/onnx-mlir
Mar 2020 - May 2022
Representation and Reference Lowering of ONNX Models in MLIR Compiler Infrastructure
Role in this project:
Back-end Developer
Back-end DeveloperContributions:152 reviews, 59 commits, 131 PRs in 2 years 2 months
Contributions summary:Gheorghe-teodor primarily contributed to the lowering of ONNX models within the MLIR compiler infrastructure, specifically focusing on the implementation of operations in the Krnl dialect. Their work involved translating ONNX operators like `MaxPoolSingleOutOp` and `Erf` to the Krnl dialect and adding related MLIR tests. The user also made improvements to the handling of memory pools and constant tensors, enabling more efficient execution and code generation.
pytorchrepresentationdeep-learningmlironnx-models
apache/camel-k
Oct 2020 - Sep 2021
Apache Camel K is a lightweight integration platform, born on Kubernetes, with serverless superpowers
Role in this project:Back-end & Automation Engineer
Back-end & Automation EngineerContributions:23 reviews, 92 commits, 18 PRs in 11 months
Contributions summary:Gheorghe-teodor primarily focused on updating and refactoring code within the `pkg/cmd/inspect.go` file, suggesting a focus on command-line interface functionality. The commits show adjustments to how dependencies are computed and outputted. Additionally, the user made several changes related to managing and building images, specifically for local builds and containerization, indicating a role in automating the build and deployment process.
integrationserverlessapache-camelopenshiftoperator
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.