GPU Engineer (AI)

Location: Remote, but candidates must already be based in Paris or elsewhere in Île-de-France. Relocation from other regions or countries is not considered.

Company

KOG builds real-time LLM inference systems for AI agents by co-designing model architecture, inference engines, and low-level GPU kernels for NVIDIA and AMD datacenter GPUs.

What you will do

• Develop and optimize GPU kernels using C++, CUDA, HIP, and low-level GPU instructions.
• Investigate GPU architecture, memory hierarchy, synchronization, execution bottlenecks, and performance issues.
• Profile and benchmark workloads to identify and validate performance improvements.
• Optimize inference operations and communication across multiple GPUs.
• Implement and optimize AI models on NVIDIA and AMD GPUs.
• Help build AI agents for automated GPU code optimization, correctness testing, benchmarking, and iterative improvement.

Requirements

• Experienced GPU Engineer able to contribute quickly and independently to complex performance problems.
• Strong hands-on experience developing and optimizing GPU kernels, ideally with CUDA C++ or HIP/ROCm.
• Deep understanding of GPU architecture, memory hierarchy, parallel execution, and performance bottlenecks.
• Strong C++ skills and experience profiling and benchmarking GPU workloads.
• Proven ownership of challenging technical projects from implementation through measured results.
• Interest in developing AI agents and tools to accelerate GPU engineering.

Nice to have

• Experience with PTX, CDNA ISA, multi-GPU systems, low-latency inference, or GPU performance portability.
• PhD in HPC, computer science, or a related field backed by substantial engineering work.
• Previous experience building AI agents is not required.

Culture & Benefits

• Full-time position within the GPU Engineering department.
• Opportunity to shape a new engineering direction focused on making model porting and GPU optimization faster and repeatable.
• Direct work with NVIDIA and AMD GPU hardware, including CUDA, PTX, HIP, and CDNA ISA.
• Quick joining availability is expected.