Autonomous AI Kernel Engineer for Frontier Inference

Etched; San Jose (CA); On-site; USD 150,000 - 225,000; Full time

Benefits offered by this job

Medical benefits

Housing subsidy

Relocation support

Wellness benefits

Daily meals

Unlimited compute budget

Job summary

Etched in San Jose (Santana Row) seeks engineers to build AI systems that autonomously translate new model architectures into production-ready kernels for Etched hardware. You will design experiments, curate datasets, and ship model-driven improvements, collaborating with architecture teams on frontier AI performance.

This role blends research with hands-on coding (Python and low-level languages) to deliver real products with measurable impact.

Qualifications

• Experience solving hard problems across stacks and domains.
• Fluency with Python and low-level code; ability to ship working solutions.
• Kernel experience and performance optimization knowledge.
• Ability to learn quickly with AI toolchains and frontier models.

Responsibilities

• Own the kernel-automation system turning architectures into verified kernels.
• Build agents to design experiments, generate implementations, and profile them.
• Design evaluations for correctness, stability, latency, and efficiency.
• Turn profiler traces and measurements into learnable signals.
• Curate datasets and optimization memories from trajectories and results.
• Ship model improvements to production and measure end-to-end impact.
• Collaborate with architecture teams to shape hardware-software roadmap.
• Evaluate new model releases and deploy the best for each stage.

Skills

Python proficiency

Low-level coding

Kernel experience

AI tooling experience

Cross-domain problem solving

Job description

Etched in San Jose (Santana Row) seeks engineers to build AI systems that autonomously translate new model architectures into production-ready kernels for Etched hardware. You will design experiments, curate datasets, and ship model-driven improvements, collaborating with architecture teams on frontier AI performance.

This role blends research with hands-on coding (Python and low-level languages) to deliver real products with measurable impact.