Чёткая практическая роль в embedded AI: работа с реальным железом и понятная передача прототипов firmware-командам. Главное ограничение - работа на месте во Франции.
Только офис во Франции
Понятные задачи
Современный стек
Масштаб продуктов
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Senior AI Engineer (Embedded Systems) (TinyML/MCUs): Prototyping neural models for audio, vision, and sensor applications that run directly on microcontrollers, with an accent on model quantisation, embedded inference, and real-device benchmarking. Focus on selecting silicon and runtimes, writing end-to-end C/C++ inference pipelines, and validating latency, memory, power, and accuracy before production handoff.
Location: France; the innovation team is based in Lille.
Salary: €55,000–€95,000 per annum.
Company
Razer develops gaming hardware and software, with a global team operating across five continents.
What you will do
Prototype embedded AI applications for audio, voice, vision, and sensor data on Razer peripherals.
Select, train, fine-tune, quantise, and optimise models for microcontrollers with and without NPUs.
Build and maintain deployment toolchains using TFLite Micro/LiteRT, ExecuTorch, CMSIS-NN, CMSIS-DSP, Vela, and vendor compilers.
Benchmark latency, memory, power, and accuracy on real silicon and development kits, comparing chips for specific use cases.
Capture camera and microphone data, check model and dataset licences, document prototypes, and hand working code and test data to firmware teams.
Requirements
5+ years of embedded software development in C/C++ on microcontrollers, including at least 2 years deploying ML models on embedded or edge devices.
Hands-on experience with Arm Cortex-M and real-time constraints, including memory budgets, DMA, interrupts, RTOS, or bare metal.
Experience with model quantisation and at least one MCU inference runtime: TFLite Micro/LiteRT or ExecuTorch.
Experience training and fine-tuning models in Python with PyTorch or TensorFlow.
Basics of audio signal processing, including filtering, FFT, and MFCC, and/or image processing.
Comfort using development kits, debuggers, logic analysers, and power measurement tools.
Ability to work independently on open problems, make fast decisions, and report results clearly.
Fluent English required.
Nice to have
Experience with Ethos-U/Vela, ST Neural-ART/ST Edge AI, NXP eIQ Neutron, or another vendor NPU toolchain.
CMSIS-NN/CMSIS-DSP optimisation, Helium/MVE, or other SIMD experience.
Small vision models or speech enhancement on constrained devices.
Zephyr RTOS experience.
Audio front ends for headsets or microphones, including AEC, beamforming, or VAD.
Published work, open-source contributions, or TinyML demos.
French language skills.
Culture & Benefits
Work on prototypes that may reach millions of Razer devices.
Join a small team with direct access to hardware, firmware, and software teams across Razer.
Collaborate with Legal on permitted data collection and with firmware teams on production handoff.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →