Senior AI Engineer

Janus Henderson is hiring a Senior Applied AI Engineer to lead delivery of major parts of the firm-wide AI transformation. The role designs, builds, evaluates, and product ionises agentic applications and platform services, then owns them through live running, including governed data access and model evaluation.

This is a senior applied AI engineering position reporting to the Principal AI Engineer. AI capability is centralized under the Head of AI, and AI Technology builds, governs, and runs the software. AI Engineering serves as the core product and platform engineering team, supporting enterprise agentic products including Nexus (agentic workspace) and Accio (centralised MCP server), alongside orchestration, evaluation, and observability services beneath Libros and PRISM, delivered with Percepta.

• Design, build, and product ionise AI products and shared platform services, taking a workstream end to end and owning it through live running.
• Build for reuse by turning proven capability into shared components, libraries, skills, tools, MCP servers, and connectors.
• Implement infrastructure as code, own CI/CD pipelines, and accept services into operation only when ownership, controls, and support are clear.
• Investigate incidents and defects in services you own, lead recovery, carry fixes through to underlying cause, and act as L3 escalation.
• Build agentic applications and workflows across agent design, prompt and context engineering, tool use, memory, retrieval, and human-in-the-loop controls.
• Implement model selection, routing, and fallback through the model gateway while handling provider change without hiding differences in capability, cost, or behaviour.
• Build orchestration layers for long-running, multi-step agents, covering permissions, workload isolation, safe execution, and dependent applications.
• Test claims for new models, frameworks, and patterns before recommending adoption.
• Build governed, AI-ready views, indexes, semantic context, and connectors over Snowflake, enterprise platforms, APIs, and external providers.
• Extend Accio so new datasets and downstream MCP servers are reachable through a single governed interface rather than one-off integrations.
• Build ingestion pipelines and data models when source data is not usable, while keeping canonical source ownership with the relevant Technology team.
• Own lineage, quality, and freshness for data that products depend on, and implement least-privilege access for users, agents, tools, and service identities.
• Own the end-to-end build of targeted business applications on the AI stack, with initial focus on Distribution and expansion to trading, investment risk, client servicing, and operations.
• Drive SaaS decommissioning for named business areas by replacing bought subscriptions with in-house builds, owning end-to-end delivery, retiring displaced products, and coordinating with Procurement and Finance to consolidate licences.
• Prioritise builds where in-house ownership improves control, reduces cost, or delivers capability not available from vendors, reusing existing Accio datasets, skills, and connectors.
• Take applications through the same evaluation, controls, and release evidence as shared products, and own support and lifecycle once live or hand over to a clear owner.
• Build evaluation and regression suites for models, prompts, agents, and platform changes, meeting agreed thresholds and instrumenting quality, safety, reliability, latency, drift, usage, and cost.
• Implement AI Governance Implementation and AI Security controls as code with secure defaults, producing release evidence proportionate to risk generated by the platform.
• Work with Percepta engineers on Libros, PRISM, and the wider estate, ensuring engineering standards and surfacing maintainability issues while they are inexpensive to resolve.
• Mentor AI Engineers through pairing, design discussion, and code review, helping establish engineering standards, agentic SDLC, and definition of done.
• Partner with AI Architecture to turn reference patterns into implementations used by teams, work with Forward Deployed Engineering so proven solutions become supported shared capability, and feed adoption data into the backlog.

• At least 6 years in software, data, or platform engineering, with a track record of shipping and operating production systems rather than prototypes.
• Production experience with LLM applications and agentic systems, including prompt and context engineering, agent development,…