The Job
Role Overview
As Staff AI Engineer in Core Engineering, you help design and build the systems behind intelligent automation, agentic workflows and large-scale retrieval at Believe. You work alongside the Agentic & Automation team (product and automation engineers, and you are the person who turns advanced AI concepts into services other teams can depend on.
This is a hands-on, deeply technical role. Most of your time goes into writing and reviewing code, designing systems, and debugging behaviour that is probabilistic rather than deterministic. The scope is enterprise-wide: what you build is consumed by engineering, product, data and business teams across the group, in around 50 countries. Reliability, architecture, quality, observability, security and cost are therefore requirements from day one, not a later hardening phase.
As a Staff engineer, your impact is measured less by what you ship alone than by what you make possible for others: the frameworks, standards and habits that let dozens of engineers build agentic features quickly and safely. You earn that influence through credibility, clear writing and teaching, not through hierarchy.
Stack: Python, Java, GCP (incl. Vertex AI), Kubernetes, Kafka & Kafka Connect, MySQL, MongoDB, Snowflake, Spark, KrakenD, n8n.
Key Responsibilities
Build our enterprise agentic framework: orchestration, retrieval and collaboration between multiple agents: tool and MCP integration, planning and hand-off, shared state, human-in-the-loop checkpoints, and graceful failure.
Write production-grade code: well-structured, efficient and testable, across production services, experimentation harnesses, and internal tooling that makes agentic capabilities easy for other functions to adopt.
Make AI systems measurable and trustworthy: golden datasets, offline and online evals, regression suites in CI, guardrails, tracing across multi-step agent runs, and telemetry on latency, quality, token spend and failure modes.
Set standards and design for security: work with platform, architecture and security on system design, internal protocols and API standards, and bring structural decisions to the Engineering Committee.
Raise the level around you: code reviews, technical discussions, design docs and workshops, mentoring, and contributing to a culture of engineering excellence.
Champion a cross-cutting technical vision and standards: help define code and architecture guidelines, guide several teams towards iterative development and autonomous decision-making, weighing the value of complex patterns against their implementation cost.
What success looks like in your first year
Squads outside Agentic & Automation ship agentic features on a framework without needing you in the room.
Every production AI capability has an eval suite, a cost profile and a visible dashboard.
The organisation has documented, adopted standards for agent design, tool exposure and retrieval, and reuses them by default.