CUDA Kernel Engineer for High-Throughput AI GPUs

Pragmatike; Cambridge (MA); On-site; USD 150,000 - 190,000; Full time

Benefits offered by this job

Sign-on bonus

Health insurance

Dental insurance

Vision insurance

401k

Equity options

Job summary

Pragmatike is seeking a CUDA Kernel Engineer to design, implement, and optimize custom CUDA kernels for NVIDIA GPUs powering large-scale AI systems. You will focus on maximizing occupancy, memory throughput, and warp efficiency while profiling workloads with Nsight and NVProf.

You’ll diagnose bottlenecks such as warp divergence and memory access patterns, and collaborate with AI systems, backend, and model acceleration teams to push GPU performance at scale.

Qualifications

• Proven track record building NVIDIA CUDA kernels from scratch.
• Strong ability to optimize kernels (tiling strategies, occupancy tuning, shared memory design, warp scheduling).
• Deep understanding of CUDA threads, warps, blocks, and grids, GPU memory hierarchy and memory coalescing, as well as warp divergence.

Responsibilities

• Design, implement, and optimize custom CUDA kernels for NVIDIA GPUs, with a focus on maximizing occupancy, memory throughput, and warp efficiency.
• Profile GPU workloads using tools such as Nsight Compute, Nsight Systems, nvprof, and CUDA‑MEMCHECK.
• Analyze and eliminate performance bottlenecks including warp divergence, uncoalesced memory access, register pressure, and PCIe transfer overhead.
• Improve GPU memory pipelines (global, shared, L2, texture memory) and ensure proper memory coalescing.
• Collaborate closely with AI systems, model acceleration, and backend distributed systems teams.
• Contribute to GPU architecture decisions, kernel libraries, and internal performance‑engineering best practices.

Skills

CUDA kernels

Kernel optimization

CUDA architecture

PCIe optimization

C++ / CUDA tooling

Tools

NCCL

NVLink

MIG

TensorRT

CUTLASS

CUDA Graphs

PTX/SASS analysis

Job description

Pragmatike is seeking a CUDA Kernel Engineer to design, implement, and optimize custom CUDA kernels for NVIDIA GPUs powering large-scale AI systems. You will focus on maximizing occupancy, memory throughput, and warp efficiency while profiling workloads with Nsight and NVProf.

You’ll diagnose bottlenecks such as warp divergence and memory access patterns, and collaborate with AI systems, backend, and model acceleration teams to push GPU performance at scale.