Gallium Defense | GPU / CUDA Architect & Engineer

Intelligent Employment; Paris; On-site; EUR 90,000 - 130,000; Full time

Job summary

Gallium Defense is seeking a GPU Architect & Engineer to own the GPU-hosted compute stack, from kernel design to memory management, and to develop and optimize CUDA functions at the heart of radar signal processing.

You will work with the SW & AI team, collaborating with RF, electronics, and systems engineering to turn demanding latency and throughput targets into robust, field-ready compute solutions for real-world testing.

Qualifications

• Master's degree in Computer Science, Electrical Engineering, or a related field, or equivalent hands‑on experience.
• 3+ years of hands‑on CUDA development with architectural judgment across a GPU pipeline.
• Deep CUDA expertise: writing, optimizing, and profiling GPU code.
• Evidence of architectural thinking on GPU‑hosted systems and AI workloads.
• Solid C++, experience with embedded Linux and real‑time constraints.
• Nice to have: radar or signal processing background; ML edge inference (TensorRT, quantization).
• Fluent in French; proficient in technical English.

Responsibilities

• GPU Software Architecture: own the GPU software structure, kernel design, memory strategy and scheduling.
• CUDA Development & Optimization: design, implement and optimize signal processing kernels for latency and throughput.
• SW / AI Integration on GPU: unify software and AI into a single performant runtime on the GPU.
• Performance, Quality & Field Support: set perf practices, write tests, review code, support field experiments.

Skills

CUDA development

GPU architecture

C++ programming

Embedded Linux

Real-time systems

French language

Technical English

Edge ML inference

Education

Master's degree in CS/EE

Tools

TensorRT

Profiling tools

CUDA toolkit

Build systems

Job description

About Gallium Defense

At Gallium Defense, they are rewriting the rules of air defense in the context of rapidly evolving drone threats. They design and build next-generation, military-grade radar systems that deliver high-performance 3D detection at competitive pricing - supporting growing demand and closing critical capacity gaps.

The Role:

As GPU / CUDA Architect & Engineer at Gallium Defense, you will report to the SW & AI Lead and be a full member of their SW & AI team. You will own the architecture of their GPU-hosted compute - from kernel design to how work is scheduled and memory is managed on the device - and develop and optimize the CUDA functions at the heart of our radar signal-processing. You'll partner closely with their SW architect to make sure that design serves the system as a whole. Hands‑on and close to the end users and the experimentation field, you will build together with the rest of the SW & AI team - which contributes alongside the other modules and teams to form a single product team - turning demanding latency and throughput targets into robust, field‑ready compute.

Key Responsibilities

• GPU Software Architecture: Take full technical ownership of how their software and AI are structured on the GPU itself - kernel design, memory strategy, scheduling and work partitioning across the device - so processing and AI run together within tight real‑time budgets. You'll work hand‑in‑hand with their SW architect on how that GPU‑side design fits the broader system.

• CUDA Development & Optimization: Design, implement, and optimize the signal processing functions - profiling and tuning for latency, throughput, and memory to meet operational requirements.

• SW / AI Integration on GPU: Bring our software and AI together into a single performant runtime on the GPU, working hand‑in‑hand with the rest of the SW & AI team and with the other modules and teams on clean interfaces.

• Performance, Quality & Field Support: Set the performance‑engineering practices (benchmarking, profiling, regression tracking), write tests, review code, and support field frequent field experiments - debugging GPU and compute issues in real environments and iterating rapidly.

Profile & Requirements

• Master's degree in Computer Science, Electrical Engineering, or a related field - or equivalent hands‑on experience.

• 3+ years of hands‑on CUDA development, with the architectural judgment to show for it - you've structured compute, memory, and data flow across a GPU pipeline (must‑have).

• Deep CUDA expertise: writing, optimizing, and profiling GPU code.

• Evidence of architectural thinking on GPU‑hosted systems - you can point to real decisions you drove on how software and AI workloads are structured to run efficiently on a GPU.

• Solid C++, experience with embedded Linux and real‑time constraints.

• Nice to have: experience with radar or signal processing, or adjacent domains (e.g. RF).

• Nice to have: experience deploying or optimizing ML models for edge inference (e.g. TensorRT, quantization, mixed precision).

• Fluent in French; proficient in technical English.

Why Join Gallium Defense?

The drone threat is growing faster than the systems designed to counter it. At Gallium, we're democratizing C-UAS radars at scale - affordable, rugged, and high‑performance - that give civil and defense operators a real answer. We think the way radar is built needs to change, and we're proving it’s possible.

• Build products that directly contribute to protecting our future

• Own critical design decisions with visible impact on the final system

• Collaborate with experts across RF, electronics, FPGA, software, and systems engineering

• Experiment directly in the field with civil and military end-users

• See your work move rapidly from concept to deployed system

• Shape both the product and engineering culture of an ambitious deep‑tech company

If you're excited about solving complex signal‑processing challenges and building systems with tangible real‑world impact, we'd love to hear from you!