Senior Staff Engineer, tem

• You own the problem as well as the solution: you've worked directly with business stakeholders to define what gets built, perhaps as a technical product lead, tech lead or engineering manager, and you don't expect a spec
• You've built and maintained production agentic systems with real consequence, including multi-agent systems and MCP servers used by real people for real work
• You have depth in RAG: you've built retrieval systems, compared approaches against requirements, and have opinions on what suits which case
• You evaluate systematically, with realistic labelled cases that include failure and permission boundaries, and you know when an agent is ready for production
• You have sound reliability and security judgement: where to enforce permissions and human approval, and how to persist task state and recover safely
• You lead agents, not people: you already orchestrate several agents in parallel, and you can review code, make architectural calls and spot where a system will fail, without coding day to day
• If you’re excited about this role but not sure you meet every requirement, we’d still love to hear from you. Your unique perspective could be exactly what we’re looking for

Desirable

• We're hiring a Senior Staff Software Engineer - Agentic to join our Agentic Service
• Our ambition is for agents to take on somewhere between 50% and 99% of the manual work people across tem do today. Teams across the company want far more agentic delivery than we can currently provide
• You'll own how agents get delivered into tem's Engines and Services. You'll work with stakeholders to understand how their work really runs, decide what to build, and make sure it's designed to be reliable and safe. You'll lead agents rather than people, running several workstreams in parallel and reviewing their output to production standards
• Nobody will hand you a spec. You'll find the problem, shape the solution and own the outcome
• In your first 12 months, you might:
• Own agent delivery into several Engines and Services at once, from first discovery conversation to measured time saved
• Set the architecture for how we use retrieval, orchestration, MCP tools, identity and evaluation across agentic work
• Run several workstreams in parallel by leading agents, holding their output to production standards
• Become the person teams want in the room when deciding what to automate
• Help teams across tem build and operate their own agents well
• Find the work: Work with Engines and Services to map how their processes really run, and decide which parts to automate, rebuild or leave alone
• Turn ambiguity into plans: Turn messy requirements into clear architecture and delivery plans that stakeholders understand and agents can execute
• Prioritise honestly: Weigh competing demands from across tem and make the trade-offs explicit before committing
• Design for reliability: Own the architecture across concurrent workstreams, with evaluation, access controls, recovery and human approval designed in from the start
• Prove the impact: Instrument what you ship so the time saved is visible, and say so when something isn't moving the number
• Raise the bar across tem: Codify recurring business logic into reusable building blocks, and help teams build and run their own agents well through design reviews, shared patterns and clear handovers

• First call with our Talent Team
• Behavioural interview with the Hiring Manager
• Technical interview with the team
• Bar Raiser interview with stakeholders