Staff AI Engineer - Agent Architecture & Behavior

Responsibilities

• Design and implement agent execution loops, planning strategies, tool interfaces, verification, and system boundaries.
• Build multi-agent delegation, coordination, context sharing, result synthesis, concurrency handling, cancellation, and stale-result management.
• Develop resilient stateful workflows with checkpoints, recovery strategies, partial-failure handling, and human intervention.
• Improve retrieval, context construction, persistent state, memory, and reusable skill representation.
• Build evaluations, analyze task trajectories, and measure quality, reliability, latency, and cost improvements.
• Evaluate models and emerging techniques; make build-versus-buy decisions and replace tools when evidence changes.
• Set engineering standards, review architecture and design decisions, and provide technical leadership as an individual contributor.
• Partner with product and infrastructure engineers on production services, integrations, secure execution, and observability.

Requirements

• Demonstrated experience building and shipping a substantial agentic system through production use or rigorous, reproducible open-source work.
• Deep practical experience with LLM tool use, planning, context engineering, evaluations, and multi-agent coordination or parallel agent/tool execution.
• Hands-on experience with browser or computer automation in an agentic system, including state observation, effect verification, and failure recovery.
• Strong software engineering skills in Python, TypeScript, or a comparable language.
• Experience with asynchronous services, state machines, persistence, concurrency, retries, and cancellation.
• Ability to distinguish model decisions from guarantees that must be enforced in code, including permissions, untrusted inputs, uncertain outcomes, and human approvals.
• Ability to design experiments, debug real system behavior, and explain measured improvements and remaining uncertainty.
• Ability to take ownership of ambiguous problems, collaborate effectively, and ship high-quality software.

Nice-to-haves

• Agent memory and retrieval
• Skill acquisition
• Reinforcement learning or post-training
• Trajectory datasets
• Sandboxed execution
• Distributed systems
• Inference optimization
• Multimodal models
• Voice models

Compensation

• Base salary: $250,000–$325,000 USD annually.
• Equity: 0.15%–0.30%.
• US visa sponsorship available.