CUDA Kernel Engineer (Remote US)
Pragmatikea day ago
Highlight
RemoteFull-timeMid
Summary
Pragmatike is seeking a CUDA Kernel Engineer to join their team remotely in the US. The role involves developing and optimizing NVIDIA CUDA kernels for high-throughput AI systems used by Fortune 500 customers.
About the Role
As a CUDA Kernel Engineer, you will design, implement, and optimize custom CUDA kernels for NVIDIA GPUs, focusing on maximizing occupancy, memory throughput, and warp efficiency. You will profile GPU workloads, analyze performance bottlenecks, and collaborate with AI systems and backend teams to enhance GPU architecture and performance.
About You
Required:
- Proven track record building NVIDIA CUDA kernels from scratch.
- Strong ability to optimize kernels including tiling strategies and warp scheduling.
- Deep understanding of CUDA threads, warps, blocks, and GPU memory hierarchy.
- Experience diagnosing PCIe bottlenecks and optimizing host-device transfers.
- Familiarity with C++, CUDA runtime APIs, and GPU debugging/profiling tools.
Preferred:
- Experience with multi-GPU or distributed GPU systems.
- Background in GPU acceleration for ML frameworks or HPC workloads.
- Knowledge of model inference optimization techniques.
- Exposure to compiler-level optimization or PTX/SASS analysis.
- Startup experience or comfort working in fast-moving environments.
Benefits
- Competitive salary & equity options
- Sign-on bonus
- Health, Dental, and Vision insurance
- 401k plan
Pragmatike
Hire the top 1% of remote tech talent with Pragmatike global reach. Trusted by remote-first companies worldwide 🧬. → Hiring process in 48 hours → 50K+ tech specialists → Operating in 60+ countries Come chat with us, whether you look for talent or new career opportunities :)
New York, NY