← jobs

Kog (https://kog.ai)

GPU Engineer | Paris, France

hn · Paris, France

25
scored

Jose is an exceptional full-stack and AI engineer with strong production systems experience, but this role requires deep expertise in GPU kernel optimization (CUDA/PTX/HIP), low-level performance tuning, and systems programming that his CV doesn't demonstrate. His strengths lie in high-level AI product development, ML model fine-tuning, and infrastructure management rather than the low-level GPU assembly and kernel-level optimization work that defines this GPU Engineer position.

Hook: Your experience shipping real-time voice inference at scale and optimizing LLM pipelines shows you understand production AI performance, but this role needs someone who has coded CUDA kernels and profiled GPU memory bandwidth.

Missing: CUDA/PTX kernel development, HIP/CDNA ISA experience, GPU assembly language proficiency, Low-level performance profiling on GPUs, Memory bandwidth optimization at kernel level, Persistent kernel design patterns, C/C++ systems programming for GPU compute

original posting ↗nicolas.constant@kog.ai

Posting

Kog (https://kog.ai) | GPU Engineer | Paris, France | REMOTE within a Europe-compatible timezone, one week per month onsite in Paris

We are hiring a GPU Engineer to work on the fastest LLM inference engine on standard datacenter GPUs.

You would own low-level kernel work in CUDA/PTX or HIP/CDNA ISA, the monokernel pipeline, profiling infrastructure inside it, scaling to the frontier MoE models that run in production, and building our own agents that optimize kernels and inference autonomously.

We generate 3,000 tokens/s per request on 8x AMD MI300X and 2,100 on 8x NVIDIA H200, at batch size 1, FP16, no speculative decoding.

At batch size 1, the decode is GEMV, so it is memory bandwidth bound, and MBU is what counts.

We rewrote the whole hot path ourselves, from the assembly on the chip up to the Transformer we designed around it, with the full decode running as a single persistent GPU kernel.

Try it at https://playground.kog.ai

Showing your code is part of the process.

If you are outside a Europe-compatible timezone, relocation to one is required.

Apply: https://jobs.ashbyhq.com/kog/e3950334-a2a6-43cc-a744-df6c38683166

Questions, email me at nicolas.constant@kog.ai