Agent Systems Engineer

About the Role

This is an early engineering hire on the Research team, focused on building the core infrastructure that powers agentic applications. You'll design and own systems that govern agent coordination, evaluation, and continuous improvement — turning stochastic research prototypes into robust, observable, production-grade platforms. The work sits at the intersection of distributed systems, machine intelligence, and product, and carries a high degree of ownership from day one.

The company operates in the AI/ML and agentic simulation space, building platforms that model goal-directed user behavior at scale for regulated, high-stakes industries such as healthcare, finance, and education. This is a hybrid, on-site role based in San Francisco, CA.

What You'll Do

• Contribute as a Member of Technical Staff within the Research team, shaping agent systems architecture from the ground up.
• Design and implement multi-agent system architectures — defining topologies, orchestration layers, tool registries, and scoped persistent state.
• Build long-running, harness-based agent platforms rather than ad-hoc LLM API integrations.
• Develop behavioral and persona models that simulate goal-directed user archetypes and enable agents to interact with real interfaces.
• Implement critique and reflection loops that produce measurable, actionable agent behavior.
• Create evaluation frameworks and prompt/reasoning schemas that measure fidelity, surface failure modes, and support both automated and human-in-the-loop (HITL) feedback.
• Transform raw agent outputs into high-signal product insights by clustering observations, scoring severity, and mapping signals to product decisions.
• Ensure production reliability through tracing, attribution, cost monitoring, fallbacks, circuit breakers, and evaluation pipelines.
• Work closely with research and product to maintain tight research–product feedback loops and iterate rapidly.

What We're Looking For

Required

• 5+ years of hands-on experience designing and implementing production-grade distributed systems and agent-based architectures (multi-agent topologies, orchestration, memory, tool registries, and execution).
• 2+ years shipping agentic LLM systems in production environments.
• Python mastery, including async programming and structured data modeling (e.g., Pydantic).
• Experience building production ML systems: tracing, cost monitoring, failure detection, and robust fallbacks.
• Experience with containerization and orchestration (Docker, Kubernetes) for large-scale agent infrastructure.
• Experience designing behavioral models, goal-directed simulations, evaluation frameworks, and HITL feedback mechanisms.
• Ability to translate agent outputs into actionable product decisions; demonstrated experience building signal and insight systems.
• Strong cross-functional communication skills; comfort collaborating across engineering, research, and product.
• High ownership mindset and comfort operating in ambiguous, frontier problem spaces.

Nice to Have

• Deep emphasis on systems over prompts: state machines, orchestration layers, and tool-use contracts for reliability and inspectability.
• Product intuition that prioritizes signal quality and decision-making impact.
• Background in simulation, behavioral research, or user modeling.

The team welcomes candidates from diverse educational and professional backgrounds.

Compensation & Benefits

• Salary: $180,000 – $250,000 USD annually
• Early-stage equity participation
• Backed at Series A/B stage with strong investor support

Note: Visa sponsorship is not available for this role.

Location

This is a hybrid, on-site position based in San Francisco, CA. The team also has a presence in London; candidates based in London with regular San Francisco availability may also be considered. Candidates must be authorized to work in the United States without sponsorship.