Senior Agentic Enablement Engineer

Description

TrueLayer is making a deliberate, CTO-led investment in agentic engineering. We are building a new team — Agentic Engineering Enablement — with a focused mission: pave the way for AI agents to succeed across TrueLayer's engineering organisation. Every bottleneck that slows an agent down is a problem to solve. Every piece of tribal knowledge locked in a Slack thread or an engineer's head is an opportunity to unlock.

As a Senior Agentic Enablement Engineer, you will be a core pillar of this team. Working closely with the CTO, you will own significant components of the agentic platform — from context engineering and tooling to evaluation systems and CI/CD — and build them into reliable, production-grade infrastructure. You will also help establish engineering norms and best practices for the team and the wider organisation.

What does good look like in Year One?

Agents routinely land coordinated changes across multiple repos and services, run e2e tests, and deploy behind guardrails without a human in the loop.

What we expect from you as a Senior Agentic Enablement Engineer:

Build and maintain the agent harness

• Write, maintain, and version AGENTS.md files, system prompt libraries, and skill definitions across the codebases and workflows that matter to TrueLayer. Treat this as infrastructure, not documentation.
• Surface tribal knowledge from across engineering and convert it into structured, agent-consumable artefacts: ADRs, runbooks, coding conventions, decision trees, and system maps.
• Design and operate MCP servers that give agents safe, real-time access to TrueLayer's internal systems, APIs, databases, CI outputs, incident history, and deployment state.
• Build and maintain TrueLayer's internal tool library: structured, versioned, testable agent capabilities that encode how we build, test, review, and operate software.

Pave golden paths

• Build reusable, opinionated golden paths, starter templates, orchestration patterns, and reference workflows, so every new agent capability is reliable and measurable by default rather than bespoke.
• Encode the safe way to do something as the easy way to do it. If the golden path is harder than building from scratch, something needs to change.
• Version and maintain these paths as products with adoption metrics and clear contracts.

Evolve CI/CD for agent-scale workloads

• Architect and operate CI/CD pipelines (GitHub Actions, Kubernetes, cloud runners) capable of handling the throughput of multi-agent workflows without becoming a bottleneck.
• Reduce CI latency through remote execution, caching, parallelisation, and test impact analysis.
• Build and maintain ephemeral, sandboxed development environments that agents can provision and tear down autonomously.
• Automate recurring manual tasks: flaky test quarantine, dependency updates, PR triage, and changelog generation.

Design the eval and oversight layer

• Design and maintain an evaluation harness: regression suites, benchmark sets, and task-specific evals for every agent capability shipped internally.
• Define and instrument agent health metrics: token spend, tool call reliability, output drift, and cost per task.
• Run structured experiments on agent configurations, prompt variants, and tooling choices — grounded in real metrics.
• Define oversight policies: human-in-the-loop checkpoints for high-risk actions, failure detection, loop detection, and hard cost ceilings. In a regulated payments environment, ungoverned agent access to production systems is a liability.
• Build guardrails and permission boundaries so agents operate with least-privilege, auditable access to internal systems by default.

Contribute to team and engineering standards

• Contribute to the architecture of the agentic platform and participate actively in key technical decisions.
• Share work, learnings, and setbacks openly. Document thoroughly. Run workshops that help engineers move from sceptics to confident users.
• Track what is working in the industry and translate it into TrueLayer-specific approaches and playbooks.

About you:

• You are genuinely AI-pilled. Not someone who has read about this: someone who has restructured their engineering workflow around AI agents and can explain exactly why, what broke, and what you fixed.
• You are actively building with AI agents — not just using AI-assisted tools occasionally, but designing and operating systems where agents do real, autonomous work. You can explain what you've built, what broke, and what you changed.
• 7+ years of software engineering with meaningful focus on infrastructure, platform engineering, or developer experience.
• Hands-on experience building agent systems: MCP servers, orchestration pipelines, tool-use schemas, AGENTS.md configurations, or evaluation harnesses for non-deterministic outputs.
• You treat prompt and context engineering as a first-class engineering discipline: precise, scoped, tested, and version-controlled.
• Solid cloud infrastructure skills: AWS preferred, with Terraform or equivalent, Docker, and Kubernetes.
• Strong written communication: able to extract complex workflows from engineers and explain architectural decisions clearly and concisely.

We would be particularly excited if you have:

• Proven experience designing and scaling CI/CD pipelines under real load — you know what slow CI costs and have fixed it.
• Experience building systems where automation or agents make coordinated changes across multiple repositories and services, run end-to-end tests, and deploy with minimal human intervention.
• Experience with agent oversight and safety as an engineering discipline: guardrails, sandboxing, permission boundaries, human-in-the-loop checkpoints, and reasoning about failure modes before they reach production.
• Familiarity with agent orchestration frameworks such as LangGraph, CrewAI, AutoGen, or equivalent custom systems.
• A background in developer experience (DevEx) and experience measuring the impact of tooling on engineering velocity in a quantifiable way.
• Experience with eval frameworks such as Braintrust, LangSmith, or custom harnesses.
• Understanding of LLM cost modelling: model selection, caching strategies, retrieval vs. context stuffing, and token budget design.
• Security or compliance experience in a regulated industry — fintech experience is a bonus.

Don't meet all the requirements?

Multiple studies have shown that women and people of underrepresented groups are sometimes less likely to apply to jobs unless they meet every single requirement. At TrueLayer, we are committed to building a diverse workforce, so if you are excited about this role and have the essential skills, but not the entire checklist - we'd still love to hear from you!

Benefits

What you can expect from us

• Competitive salary and meaningful equity in the company 💰
• Flexible hours and hybrid working - we offer a hybrid approach to work of 3 days per week remote working 🏡 and 2 days per week in our new offices in London 🇬🇧 Milan 🇮🇹 and Dublin 🇮🇪
• Need to collect the kids from childcare? Love a workout in the gym first thing? No worries, we trust you to do your best work within our hybrid framework 🧘
• Flexible holiday policy, with 24 days as standard ✈️
• 2 volunteering days to support causes important to you 🌳
• Generous parental leave, above and beyond statutory requirements and with no minimum tenure 👨‍👩‍👧
• Meal vouchers for every day you’re working, including while remote
• Budget to spend on learning & development each year 📚
• 12 fully-paid wellbeing days a year and your birthday off (on top of the holiday allowance) 🕊
• Membership of mental wellbeing platform Spill 🧘🏻‍♀️
• CCNL commercio (contratto collettivo nazionale di lavoro)

Compensation

€85,000 - €110,000/yr + equity.

About our hiring process

Our hiring process typically consists of four stages:

1. A call with one of our Talent Acquisition Partners.
2. A take-home exercise
3. 2-hour video interview with two TrueLayer engineers.
4. 1-hour total final round video interviews

Full details on the interview process, including preparation tips and the qualities we look for in future TrueLayeristas, can be found on our blog