Hubert De La Jonquière
Inference Team Lead at H Company
Greater Paris Metropolitan Region France
Summary
🤩
Rockstar🎓
Top SchoolHubert De La Jonquière is an experienced inference and ML engineering leader with a decade of practice building and shipping on-device voice and signal-processing systems. Currently leading an inference team in the Greater Paris area, he has driven Sonos’s on-device voice stack across CPU, NPU, DSP and MCU targets and managed distributed software and ML teams in the US and France. His background spans embedded bare-metal inference, production ASR, wake-word engines and memory-efficient keyword spotting—work that culminated in a Sonos patent and product launches on multiple hardware generations. Hubert is also an active open-source contributor: he extended Sonos’s tract inference library to support complex-number tensor types, removing dependencies and resolving build issues to broaden ONNX/TensorFlow inference capabilities. He combines deep applied math and engineering training from Centrale Paris and Imperial College with hands-on systems craftsmanship, making him as comfortable optimizing low-level deployments as shaping team strategy.
10 years of coding experience
10 years of employment as a software developer
Master of Science (MSc) in Computing (Specialism), double degree with Ecole Centrale Paris, Software Engineering, Master of Science (MSc) in Computing (Specialism), double degree with Ecole Centrale Paris, Software Engineering at Imperial College LondonMaster of Science (MSc) in Engineering, Master of Science (MSc) in Engineering at Ecole Centrale ParisMathematics and Physics, Mathematics and Physics at Lycée CharlemagneLycée Saint Jean de PassyBachelor's degree in Mathematics for Modelisation and Decision, double degree with Centrale Paris, Applied Mathematics, Bachelor's degree in Mathematics for Modelisation and Decision, double degree with Centrale Paris, Applied Mathematics at Université Paris Dauphine
Github Skills (9)
rust-lang10
data-structures10
rust10
onnx10
tensorflow10
complex-numbers10
data-structure10
neural-network9
artificial-intelligence8
Programming languages (4)
Github contributions (5)
sonos/tract
Sep 2021 - Nov 2022
Tiny, no-nonsense, self-contained, Tensorflow and ONNX inference
Role in this project:
Back-end Developer
Back-end DeveloperContributions:15 reviews, 54 commits, 44 PRs in 1 year 3 months
Contributions summary:Hubert significantly contributed to the `tract` repository by adding support for complex numbers within the Tensor data structure. This involved implementing new data types (ComplexI16, ComplexF16, etc.), modifying macros to dispatch operations for complex numbers, and updating the `DatumType` enum to include these new types. Furthermore, the user removed dependencies and fixed compilation issues related to these changes, enhancing the library's capabilities within the context of Tensorflow and ONNX inference.
rust-libraryno-nonsenseself-containedrustdeep-learning
sonos/rustling
Apr 2017 - Jul 2018
Rust implementation of Duckling
Contributions:14 releases, 129 commits, 30 PRs in 1 year 3 months
rustrust-langduckling
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.