GPU Kernel Engineer – CUDA, Triton & Accelerator Performance

About the role

Anyone AI is seeking experienced GPU Kernel Engineers to work on a high‑impact project that reviews, debugs, and evaluates high‑performance compute kernels used in AI workloads. The role is remote and part‑time, ideal for engineers who enjoy low‑level hardware optimization.

Key responsibilities

• Implement, debug, and optimize GPU and accelerator kernels using CUDA, Triton, NKI, Pallas, and related frameworks.
• Translate kernels between frameworks, migrate them across hardware platforms, and verify numerical correctness.
• Profile performance with tools such as Nsight, NCU, and roofline analysis, and identify bottlenecks.
• Assess kernel benchmarks, evaluate tolerance thresholds, and provide actionable feedback.

Required profile

• 3+ years of hands‑on experience developing, optimizing, or debugging GPU or accelerator kernels.
• Strong understanding of GPU performance metrics, memory hierarchy, occupancy, and floating‑point tolerance.
• Ability to distinguish software defects from environment or configuration issues.

Required skills

• CUDA
• Triton
• NKI / AWS Neuron
• Pallas / JAX
• Nsight, NCU, roofline analysis
• Kernel profiling and benchmarking
• Memory bandwidth and compute throughput optimization