MTS - Kernel Engineer (AI)

Location: Mountain View, California, United States

Salary: $260,000–$320,000 per year

Company

hirify.global is hiring for an early-stage AI hardware company developing a custom accelerator platform.

What you will do

• Develop and optimize compute kernels for a custom AI accelerator.
• Implement tensor operations, data-movement patterns, and memory-hierarchy optimizations.
• Build profiling infrastructure to measure throughput, latency, and performance against architectural targets.
• Define execution, synchronization, register-passing, and memory-management strategies across control cores and specialized compute engines.
• Enable kernel execution in simulation and pre-silicon environments and collaborate with compiler, architecture, and systems teams.
• Investigate bottlenecks, recommend hardware or software improvements, and create technical documentation and development guides.

Requirements

• Strong C and C++ programming skills in production-quality systems or performance-critical code.
• Deep experience with CUDA or a comparable accelerator-programming model.
• Strong knowledge of warp, wavefront, or thread-group execution, memory coalescing, shared or local memory, registers, caches, synchronization, and data locality.
• Ability to reason about pipelines, execution units, memory hierarchies, bandwidth, latency, and data-movement costs.
• Experience profiling and optimizing performance, including measuring throughput and latency and identifying bottlenecks.
• Practical understanding of GEMM, convolution, attention, reductions, scatter/gather, and elementwise operations, plus Python for tooling and automation.

Nice to have

• Experience with Triton, CUTLASS, MLIR, LLVM, kernel DSLs, or compiler infrastructure.
• Knowledge of RISC-V, x86, ARM64, architectural simulators, or instruction-set simulators.
• Experience with custom ASICs, GPUs, NPUs, FPGA development, RTL, Verilog, or SystemVerilog.
• Background in high-performance or scientific computing and hardware-software co-design.

Culture & Benefits

• Cross-functional collaboration with architecture, compiler, simulation, hardware, and systems teams.
• Direct influence on the kernel programming model and the efficiency of custom accelerator hardware.