Applied AI Engineer, Etched

• A track record of solving hard problems across stacks and domains—you enjoy being dropped into unfamiliar territory and figuring it out
• Hands-on experience building and shipping LLM-based agents or AI tooling that people depend on: context engineering, tool integration, orchestration, evaluation, and failure analysis
• Strong software engineering skills, especially in Python. You can build reliable integrations with complex engineering tools, debug unfamiliar systems, and direct AI to write code well. We do not care whether you write code from scratch—we care whether you ship things that work
• Interest, experience, or an academic exposure to in electrical or mechanical engineering, or a record of learning a technical domain deeply enough to build useful tools for its practitioners. You can reason about physical constraints and distinguish a plausible design from a verified one
• Fluency using AI to learn and ramp on new problems—agentic coding tools, deep research, and frontier models are how you work, not an add-on
• An eval-driven mindset: you measure whether AI systems work, investigate where they fail, and use those failures to improve them
• Comfort moving between research exploration, agentic experimentation, engineering-tool debugging, and production execution

Desirable

• Designing frontier AI hardware means solving electrical and mechanical problems together: schematics and board layouts, power delivery and signal integrity, packaging and cooling, manufacturing and assembly
• Every decision changes the constraints on the next
• Future breakthroughs will come from AI systems that can reason across these domains, operate engineering tools, run thousands of experiments, and learn which designs work in the physical world
• You will build those systems
• Your mandate is to build agents that take electrical and mechanical designs from requirements through verified, manufacturable outputs
• These agents should create and modify CAD models, design schematics and PCB layouts, run simulations, diagnose failures, and iterate with our engineers, pushing the limits of how much of the design process AI can own end to end
• Etched offers a uniquely tight research loop: chips, boards, racks, cooling systems, manufacturing, and dedicated in-office compute under one roof
• By architecting and orchestrating agent loops, you will turn proprietary designs, simulation results, engineering decisions, and physical test data into systems that get better with every experiment—and ship the hardware they help design
• Build and own AI systems that turn engineering requirements into verified electrical and mechanical designs, carrying intent and constraints from initial concepts through manufacturing outputs
• Build agents that operate CAD, EDA, and simulation tools to generate designs, run experiments, inspect results, diagnose problems, and iterate with our teams, to the limits of model autonomy
• Develop workflows spanning component selection, schematic capture, PCB layout, mechanical CAD, assemblies, thermal analysis, and electrical and structural simulation
• Build the tool integrations and representations agents need to reason about geometry, connectivity, materials, tolerances, and coupled electrical, mechanical, thermal, and manufacturing constraints
• Design evals that measure engineering correctness, constraint satisfaction, simulation accuracy, manufacturability, and performance on real design tasks
• Turn simulation outputs, design-rule checks, engineering reviews, and physical measurements into structured feedback models can learn from
• Curate proprietary datasets and design memory from complete trajectories, expert demonstrations, failed approaches, and manufactured outcomes
• Build reproducible experiment infrastructure so design revisions, tool actions, simulation settings, and results remain traceable and experiments can run at scale
• Ship agent-generated designs with our electrical, mechanical, and manufacturing teams, and quantify improvements in design cycle time, hardware performance, and engineering effort
• Continuously evaluate new model releases and deploy the best models and methods for each stage of the design loop