Applied AI Engineer, Kernel Performance

Etched
San Jose
Workplace: OnsiteFull timeUSD 150,000 - 225,000 annuallyFunction: Data Science & Machine LearningSkills: ["Problem-solving","High agency","Comfort with ambiguity","Eval-driven mindset","Collaboration"]

Build AI systems that autonomously convert newly released model architectures into correct, production-ready kernel implementations optimized for Etched hardware. You’ll design agents for experiment generation (compile, profile, diagnose, iterate), create evals for correctness and efficiency, and turn profiler and hardware signals into learnable data. Partner with other architecture teams to shape abstractions, deploy the best optimizations, and quantify end-to-end performance impact.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
Etched
Etched
1 month ago

Applied AI Engineer, Kernel Performance

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 9 hours agoStatus: Live

Job Summary

Build AI systems that autonomously convert newly released model architectures into correct, production-ready kernel implementations optimized for Etched hardware. You’ll design agents for experiment generation (compile, profile, diagnose, iterate), create evals for correctness and efficiency, and turn profiler and hardware signals into learnable data. Partner with other architecture teams to shape abstractions, deploy the best optimizations, and quantify end-to-end performance impact.
Location: San Jose
Workplace: Onsite
Employment Type: Full time
Job Function: Data Science & Machine Learning

Key Responsibilities

  • •Own the system that turns new model architectures into verified, production-ready kernels and model mappings.
  • •Build agents that understand Etched hardware, design experiments, generate implementations, compile and profile them, diagnose bottlenecks, and iterate.
  • •Design evals covering correctness, numerical stability, latency, and efficiency.
  • •Convert profiler traces, simulation, hardware counters, and expert judgment into structured signals models can learn from.
  • •Ship model-generated improvements to production, quantify end-to-end performance impact, and partner with other architecture teams to shape abstractions and the hardware-software roadmap.

Pay and Benefits

Salary: USD 150,000 - 225,000 annually
Perks:Health InsuranceHousing SubsidyRelocation SupportMeal AllowanceUnlimited Compute

Key Requirements

  • •Comfort with both Python and low-level code, including reading, modifying, debugging, and directing AI to write working implementations.
  • •Kernel experience: you’ve written or tuned kernels and can explain optimization mechanisms and their performance impact.
  • •Fluency using AI to learn and ramp on new problems, including agentic coding tools, deep research, and frontier models.
  • •Moving fluidly between research exploration, agentic experimentation, low-level debugging, and production execution.
  • •A track record of solving hard problems across stacks and domains, with the ability to figure out unfamiliar problems quickly.
Experience:HardwareAIInferenceKernelsLLMsLLM-based agents
Skills:Problem-solvingHigh agencyComfort with ambiguityEval-driven mindsetCollaboration
Tech Stack:PythonLLMsRAGTransformersTransformer-like architecturesCompilerProfilerHardware countersSimulation

Company Brief

Etched
Designs transformer‑specialized AI inference ASICs (product: Sohu) to accelerate large‑language‑model workloads, working with TSMC for fabrication and targeting energy‑efficient inference performance.
Industry: Hardware Devices
Company Size: Medium (51 to 250 employees)
Growth: Early Stage Startup
Valuation: Unicorn (USD 1B+)
Funding: Series A
Headquarters: San Jose, United States
Founded: 2022
WebsiteLinkedInGlassdoor