Senior Engineer, Kernel Development and Optimization

Job Title

Senior Engineer, Kernel Development and Optimization

Role Summary

Work on the Kernel Development and Optimization team to implement and optimize performance-critical kernels for ML and HPC workloads on Tenstorrent accelerator hardware. The role focuses on producing measurable performance improvements across kernels, runtimes, and host orchestration.

This position is hybrid, based out of Belgrade, Serbia.

Experience Level

Senior. The posting indicates level will be assessed during interview and offers will align with the assessed level; no specific years of experience are provided.

Responsibilities

Deliver high-performance kernels, own performance from diagnosis to production, and provide tooling and tests to maintain gains.

• Design, implement, and optimize GPU-style kernels (matrix multiplication, attention primitives, data-movement operations).
• Identify performance bottlenecks using profiling and benchmarking; deliver measurable throughput and latency improvements.
• Contribute to host-side orchestration code and parallelization strategies.
• Develop micro-benchmarks, regression tests, and tooling to ensure correctness and sustained performance.
• Debug complex runtime and kernel-level issues in large codebases.
• Collaborate closely with compiler, runtime, ML, and hardware teams to integrate kernels into production systems.

Requirements

Must-have technical skills are listed first; a few desirable skills follow.

• Must-have: Strong C++ systems programming experience, especially writing performance-critical or low-level software.
• Must-have: Proven use of profiling and benchmarking to guide optimization decisions.
• Must-have: Solid understanding of concurrency, synchronization, memory hierarchy, and latency-hiding techniques.
• Must-have: Experience debugging complex runtime or kernel-level issues in large codebases.
• Must-have: Experience designing or optimizing ML/HPC kernels (e.g., matrix multiply, attention, data movement).
• Nice-to-have: Experience with GPU/CUDA ecosystems, compiler or runtime development, or accelerator hardware (RISC-V or custom architectures).
• Nice-to-have: Familiarity with micro-benchmarking frameworks, regression testing for performance, and AI-assisted kernel tooling.

Education Requirements

Not specified.

About the Company

Company: Tenstorrent

Headquarters: Austin, Texas, United States

Tenstorrent is a technology company focused on designing innovative computing solutions. They are known for their expertise in the development of advanced hardware, including ASICs and SoCs, aimed at enhancing performance and efficiency in various applications.