Theodoros Theodoridis
Research Assistant at ETH Zürich
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolTheodoros Theodoridis is a research-focused software engineer with 12 years of experience, currently working as a Research Assistant at ETH Zürich and pursuing advanced graduate studies in computer science. He combines strong C++ systems expertise with practical machine learning tooling experience, contributing to high-profile open-source projects like Facebook Research’s TensorComprehensions and the Unvanquished game engine. His contributions span audio subsystem engineering—designing codecs and integrating OpenAL—and improving autotuning, caching, and robustness in ML compilers, demonstrating attention to both low-level performance and reproducibility. Based in Zurich, he bridges academic research and production-grade engineering, with a knack for simplifying complex subsystems (e.g., replacing class hierarchies with lean data structures) to improve maintainability and performance.
12 years of coding experience
Doctor of Science, Computer Science, Doctor of Science, Computer Science at ETH Zürich
Github Skills (15)
auto-tuning10
audio10
ogg10
machine-learning10
wav10
c-language10
opus10
game-development10
cpp10
cprogramming-language10
openal9
id39
cuda9
caching8
code-optimization8
Programming languages (7)
Github contributions (5)
facebookresearch/TensorComprehensions
Feb 2018 - Jul 2018
A domain specific language to express machine learning workloads.
Role in this project:
ML Engineer
ML EngineerContributions:53 commits, 21 PRs, 85 pushes in 4 months
Contributions summary:Theodoros primarily contributed to the Tensor Comprehensions project by addressing bugs and improving the autotuning capabilities. Their work included fixing an email address in documentation, correcting an issue in the genetic tuner related to signal handling, and refactoring caching mechanisms to utilize canonicalized tensor descriptions. Furthermore, the user updated the autotuner's behavior to properly handle user-defined parameter sizes, ensuring these configurations are included in the tuning process, and corrected a bug in the compilation cache.
workloadsexpressmachine-learningdomaindomain-specific
Unvanquished/Unvanquished
Mar 2014 - Apr 2014
An FPS/RTS hybrid game powered by the Daemon engine (a combination of ioq3 and XreaL)
Role in this project:Back-end Developer
Back-end DeveloperContributions:21 commits in 21 days
Contributions summary:Theodoros primarily contributed to the audio subsystem of the game, focusing on audio file loading and processing. They implemented a hierarchy of audio file classes (WAV, OGG, Opus) with associated codecs, utilizing vorbisfile and opusfile libraries for decoding. The user also refactored the audio system, replacing the audio file hierarchy with a simpler AudioData struct and integrating it with OpenAL. They demonstrated proficiency in C++ and sound file formats.
gamexreallibrocketrtstremulous
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.