Staff Engineer

We’re looking for a Staff Engineer to help build Culture Amp’s internal facing Agent Hub, the central technical capability for designing, shipping, and governing autonomous agents across the business.

This is genuinely greenfield. You’ll help define the architecture, engineering patterns, operating model, and guardrails that make agentic systems useful, secure, observable, and safe to run in production.

You’ll work across the wider business to turn ambiguous friction into practical solutions, while creating the platform and patterns that help teams build and operate AI-enabled workflows independently.

You’ll be one of the founding members of a small high-ownership team building Culture Amp’s agentic capability from the ground up. The Hub works as a central enablement function: setting standards, building shared capabilities, and helping teams across the business move from experimentation to governed, production-ready agents and workflows.

You’ll work closely with AI champions across the organisation. The team’s success depends on strong technical judgment, genuine collaboration, and a bias toward shipping solutions that create real value.

• Play a lead role in setting the technical direction for the Agent Hub’s foundational capabilities, including agent execution, orchestration, MCP tools, identity, context and knowledge, evaluation, and operational visibility.
• Design, build, and operate autonomous agents, workflows, and integrations for internal use cases — choosing thoughtfully between deterministic automation, simple tools, single agents, and multi-agent systems.
• Build reusable MCP servers and tool patterns, including clear schemas, credential brokering, identity-aware access, per-call authorisation, and auditable interactions with systems of record.
• Shape the knowledge and context layer that grounds agents in Culture Amp’s organisation-specific information, using appropriate combinations of document retrieval, vector search, knowledge graphs, registries, and semantic models.
• Establish the engineering practices that move solutions from laptop-based experimentation to managed services: secure deployment, testing, evaluation, observability, cost awareness, reliability, and safe human escalation.
• Partner with teams across the business to identify high-value problems, shape loosely defined opportunities, and deliver solutions that people actually adopt.
• Apply and evolve Culture Amp’s agent registry, risk framework, and governance patterns so every deployed agent has clear ownership, purpose, authority, data access, and accountability.
• Raise the capability of the wider function through coaching, architecture guidance, reusable patterns, documentation, and playbooks — enabling teams to move quickly within well-designed guardrails.

• Strong backend engineering skills (ideally in TypeScript and/or Python), including APIs, testing, service design, and building production systems or integrations end to end.
• Hands-on experience delivering and operating cloud-native solutions in AWS, with a strong understanding of security, identity, permissions, secrets, and reliability in production.
• Practical experience with AI agents, Claude or similar tooling, MCP or comparable integration patterns, and orchestration across single-agent, multi-agent, workflow, and human-in-the-loop systems.
• Working knowledge of agent platforms and runtime concerns such as tool permissions, identity propagation, session state, token and time budgets, retries, evaluation, tracing, cost attribution, and auditability.
• Experience designing knowledge or memory architectures, including vector or graph-based retrieval, and the judgment to choose between build, buy, configure, workflow, and agentic approaches.
• A track record of taking ambiguous, unscoped problems from discovery through to a shipped and governed solution, earning stakeholder influence through delivery.
• The ability to define platform boundaries and self-service patterns so other teams can move independently without compromising security or operational quality.

• Energised by building something from zero — you’d rather define the pattern than inherit one.
• Comfortable holding ambiguity, making decisions with incomplete information, and iterating in the open rather than waiting for a perfect plan.
• A systems thinker who can move between architecture, implementation, operations, and the human behaviours needed for adoption.
• Pragmatic about hype: excited by autonomous agents, MCP, and knowledge graphs, but focused on what creates measurable value and can be operated safely at scale.
• A generous, team-first collaborator who shares context freely, lifts the people around you, and treats the Hub’s success as more important than individual credit.