Lead Agent Research for Frontier AI Systems

Etched; San Jose (CA); On-site; USD 175,000 - 275,000; Full time

Benefits offered by this job

Medical, dental, and vision packages

Relocation support

Housing subsidy

Daily lunch and dinner

Unlimited compute budget

Wellness benefits

Job summary

Etched is building hardware for frontier intelligence and leads in inference-focused systems. As a research lead, you will guide the direction for self-improving agents and spearhead experiments that reliably complete long-running engineering tasks, balancing memory, context, tooling, and training data.

You will work across ASIC development, inference, and supercomputing with engineers, writing code, running experiments, and taking ideas to production in a hands-on leadership role.

Qualifications

• Demonstrated ability to turn ambiguous problems into clear hypotheses and experiments.
• Strong engineering ability across research exploration, experimentation, debugging and production.
• Experience building agents for complex, multi-step tasks with context, memory and tools.
• Hands-on experience with LLM post-training including SFT and RL.
• Ability to deliver meaningful progress under compute and data constraints.

Responsibilities

• Own the research agenda for agentic systems across the organization and set priorities.
• Work with domain engineers to translate failures into research questions and evaluations.
• Develop long-running task systems with memory, context management, and tool use.
• Design evaluations to show real gains transfer to production.
• Train models with SFT, RL, and distillation within the training loop.
• Scale experiments across many agents while controlling compute costs.

Skills

Research taste

Engineering execution

Agent experimentation

Context, memory, tools

LLM post-training

High agency

Technical direction

Production code practices

Tools

SFT

RL

Distillation

Job description

Etched is building hardware for frontier intelligence and leads in inference-focused systems. As a research lead, you will guide the direction for self-improving agents and spearhead experiments that reliably complete long-running engineering tasks, balancing memory, context, tooling, and training data.

You will work across ASIC development, inference, and supercomputing with engineers, writing code, running experiments, and taking ideas to production in a hands-on leadership role.