Agent Systems Engineer
About this role
This role sits on Seldon's Research Tribe as an early engineering hire shaping the core systems that power agentic applications; you’ll build the infrastructure that governs agent coordination, evaluation, and continuous improvement to make stochastic models reliable in production. The work bridges distributed systems, machine intelligence, and product to turn research prototypes into robust, observable systems that deliver measurable behavior.
What youll do
- Contribute as a Member of Technical Staff focused on Agent Systems within the Research Tribe.
- Work from Seldon’s San Francisco office (hybrid) or London with regular SF presence as specified.
- Design and implement multi-agent systems architecture that defines topologies, orchestration layers, and scoped persistent state.
- Build long-running, harness-based agent platforms rather than ad-hoc LLM API wiring.
- Develop behavioral and persona models that simulate goal-directed user archetypes and enable agents to interact with real interfaces.
- Implement critique and reflection loops to produce measurable, actionable agent behavior.
- Create evaluation frameworks and prompt/reasoning schemas that measure fidelity, failure modes, and support automated and HITL feedback.
- Transform raw agent outputs into high-signal 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.
What Seldon is looking for
- 5+ years building complex software and 2+ years shipping agentic LLM systems in production.
- Deep experience designing orchestration systems, sandboxed execution, and multi-agent coordination at scale.
- Strong emphasis on systems over prompts: state machines, orchestration layers, and tool-use contracts for reliability and inspectability.
- Python mastery, including async programming and structured data modeling (e.g., Pydantic).
- Proven track record of shipping iteratively with close research–product loops and high ownership.
- Comfort with ambiguity and shaping new best practices at the frontier of agent systems.
- Product intuition that prioritizes signal and decision-making impact.
- Open to candidates from diverse educational and professional backgrounds.
Company at a glance
Seldon provides decision layer for product development using a multi-agent simulation engine to predict user responses and surface friction. It uses personas from analytics and ethnography, matches behavior 87% across Big Five traits, and is SOC 2 Type II compliant.
Top Benefits
- WFH: 2 days/week (Wed & Fri) to catch up on laundry, save time commuting, and let your cat walk across the keyboard
- Adventures: 2 annual offsites (we’re back from Las Vegas and planning something fun in Sept), plus we host regular team dinners and events in the city
- Environment: Beautiful office in SoMa, fully stocked, and optimized for deep work
- PTO: 28 days (inc 5 company days on stats) that we all take (even our founders)
- Health: 100% paid medical, dental & vision
What happens next
Skip the application pile. I get you in front of the people who decide.
Confirm the fit
A few questions to make sure this role is the right shape for you. Two minutes.
I pitch you to the company
I write the intro, send it to the founder, and handle the back-and-forth.
A meeting lands on your calendar
When the company wants to meet, I get the call on your calendar. You just show up.
Know someone who'd be great for this?
