Eyyüb Sari
Machine Learning Engineer at Meta
Montreal, Quebec, Canada
Summary
👤
Senior🎓
Top SchoolEyyüb Sari is a Machine Learning Engineer with 12 years of experience, currently advancing ML infrastructure and model deployment at Meta from Montreal. He specializes in deep learning model compression—quantization, pruning, and knowledge distillation—with a track record of shipping integer-only and low-bit solutions for on-device and cloud products at Huawei and in research. He built a PyTorch-based internal compression framework adopted across product teams, patented and published novel pruning methods, and delivered production-ready compressed face recognition and ASR models. Equally comfortable in systems and research, he implemented integer-only LSTM and an accelerated on-device training prototype using C++/ARM NEON, demonstrating both algorithmic rigor and low-level optimization skills. An active open-source contributor, he has contributed parser and type-inference improvements to the widely used facebook/flow project, reflecting a breadth that spans ML, backend engineering, and developer tooling. Colleagues describe him as a research-driven engineer who turns cutting-edge compression research into practical, product-ready implementations.
12 years of coding experience
5 years of employment as a software developer
EpitechBac S - spé ISN, Bac S - spé ISN at Lycée Joffre - MontpellierComputer Science, Computer Science at Dublin City University
French, English
Stackoverflow
31reputation
592reached
1answer
0questions
Github Skills (12)
flowtype10
type-system10
work-flow10
javascript10
parser10
type-inference10
flow-control10
parsing10
parse10
control-flow10
configuration-management8
testing7
Programming languages (7)
Github contributions (5)
facebook/flow
Jul 2015 - Dec 2015
Adds static typing to JavaScript to improve developer productivity and code quality.
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:8 commits, 16 PRs, 58 comments in 5 months
Contributions summary:Eyyüb contributed to the `facebook/flow` repository by implementing and refining features related to static typing in JavaScript. Their work included adding `declare type` support within modules, improving the parser, and enhancing the type inference process. They also addressed issues related to uninitialized variable declarations and file extension configuration, enhancing the robustness of the type checking system. Further contributions involved allowing `null` type annotations and refining the handling of declared modules with exports.
static-typingproductivityjavascriptdeveloper-productivitydataflow-programming
oklm-wsh/ppx_tyxml
Nov 2014 - Aug 2015
Contributions:50 commits, 8 pushes in 9 months
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.