GPU Kernel Engineer – CUDA, Triton & Accelerator Performance
Anyone AI · Remote · Argentina - Fully Remote · Uruguay · Chile · Ecuador - Fully Remote · Portugal · Mexico - Fully Remote · Colombia - Fully Remote · Spain · Brazil ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote
- Part-time contract
- CUDA & Triton
- GPU performance
- Kernel debugging
Anyone AI seeks experienced engineers to evaluate high-performance compute kernels used in artificial intelligence. You will assess technical correctness, efficiency, and reproducibility of implementations across various frameworks.
The position involves deep-dive debugging, performance profiling, and providing actionable feedback on kernel optimizations and hardware migrations.
What you'll do
- Review GPU and accelerator kernel implementations for correctness
- Compare outputs against reference implementations
- Evaluate numerical tolerance thresholds and fairness of benchmarks
- Identify performance bottlenecks and optimization opportunities
- Assess realism of performance targets against hardware limits
- Review kernel translations and hardware migration strategies
- Diagnose compilation, driver, memory, and runtime issues
- Distinguish between software defects and configuration errors
- Provide clear, actionable technical feedback to teams
What you'll need
- 3+ years developing, optimizing, or debugging GPU kernels
- Strong experience with at least two: CUDA, Triton, NKI, Pallas
- Deep understanding of GPU performance optimization techniques
- Proficiency with profiling tools like Nsight, NCU, or roofline analysis
- Knowledge of memory bandwidth, compute throughput, and occupancy
- Understanding of shared memory, register pressure, and bank conflicts
- Strong grasp of floating-point numerical correctness and tolerances
- Experience debugging kernel compilation and runtime issues
- Ability to identify environment problems versus genuine optimization challenges
Nice to have
- Experience with NVIDIA GPUs and custom accelerator ecosystems
- Familiarity with AWS Trainium, TPU, JAX, or other accelerators
- Compiler engineering experience with MLIR or XLA
- Contributions to GPU or ML kernel libraries
- Experience with cuBLAS, cuDNN, or Triton community kernels
- Background in AI model evaluation, RLHF, or benchmark development
Stack
- CUDA
- Triton
- NKI
- Pallas
- JAX
- AWS Neuron
- Nsight
- NCU
- MLIR
- XLA
- cuBLAS
- cuDNN
How you'll work
- Fully remote position
- Part-time engagement
- Project-based consulting role