Your opportunityJanus Henderson is undertaking a firm-wide AI transformation to become the most technologically sophisticated asset manager in the industry. Our AI capability sits in a single centralised function under the Head of AI, and AI Technology is the part of it that builds, governs, and runs the software.
AI Engineering is the core product and platform engineering team within it. Where Forward Deployed Engineering embeds with a business unit to solve one team's problem, AI Engineering builds the enterprise products every other team depends on: Nexus, our agentic workspace, where employees build, test, run, and manage governed AI applications, agents, and shared skills; Accio, our centralised MCP server, which acts as a passthrough and logic centre for enterprise datasets, consuming other MCP servers and presenting them through one governed interface; and the orchestration, evaluation, and observability services beneath Libros and PRISM, two of the projects we are delivering with Percepta.
As a Senior Applied AI Engineer, reporting to the Principal AI Engineer, you will lead the delivery of major parts of that estate - designing and building agentic applications and platform services, taking them through evaluation and our AI governance checkpoints into production, then owning them once live. You will set implementation patterns alongside AI Architecture, review other engineers' work, and be the person the team turns to when a production system behaves in a way nobody predicted. You will also work beside Percepta's engineers, making sure each platform enters service with a design the team understands and more than one person able to extend it.
You will also own the end-to-end build of business applications for specific parts of the firm - not only the horizontal products every team shares, but targeted applications built on the AI stack for a named business area, carried from first requirement through to a supported production service. Distribution is the first focus. Much of this work is deliberate SaaS decommissioning: where a capability currently bought as a SaaS subscription can be built in-house instead, you own that build end-to-end and see the displaced product retired, so the firm consolidates onto governed, in-house capability rather than paying for overlapping tools. You will take the same approach into other business areas - trading, investment risk, client servicing, and operations among them - choosing the builds where owning the application gives the firm more control, lower cost, or capability no vendor sells.
Your toolkit spans Python and SQL, LLMs and agent frameworks, MCP, Azure AI Foundry through the AI Team's model gateway, Snowflake and Microsoft Fabric, and Azure with Terraform, Docker, and CI/CD. AI Engineering is being established for the first time here, so you will shape how the team works rather than inherit a settled routine.
What success looks like- The products and services you own are in production, used by real teams, and operate with clear ownership, telemetry, and evaluation results.
- What you build gets reused. Other engineers, agents, and applications consume it through Accio or as a skill, tool, or connector on Nexus rather than building their own.
- Platforms built jointly with Percepta transfer into our ownership with maintainable designs and a team that can extend them without external help.
- Evaluation, observability, and control are built into what you ship rather than bolted on before release.
Your responsibilitiesBuild and run our AI products and platform- Design, build, and productionise AI products and shared platform services, taking a workstream end to end and owning it through to live running.
- Build for reuse, turning capability proven in one product 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 the services you own, lead recovery, carry fixes through to the underlying cause, and act as L3 escalation.
Engineer agents, orchestration, and the application layer- Build agentic applications and workflows covering 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, handling provider change without hiding differences in capability, cost, or behaviour.
- Build the orchestration layer that long-running and multi-step agents depend on, covering permissions, workload isolation, safe execution, and the applications built on it.
- Test the claims made for new models, frameworks, and patterns before recommending we adopt them.
Make enterprise data AI-ready- 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 one governed interface rather than one-off integrations.
- Build ingestion pipelines and data models where source data does not arrive usable, while canonical source ownership stays with the relevant Technology team.
- Own lineage, quality, and freshness for the data your products depend on, and implement least-privilege access for users, agents, tools, and service identities.
Build business-area applications and decommission SaaS- Own the end-to-end build of targeted business applications on the AI stack - Distribution first, then areas such as trading, investment risk, client servicing, and operations - delivered for a named business owner rather than for every team at once.
- Drive SaaS decommissioning: where a bought subscription can be replaced by an in-house build, own the delivery of that replacement end-to-end, see the displaced product retired, and work with Procurement and Finance to consolidate the licence.
- Prioritise the builds where in-house ownership gives the firm more control, lower cost, or capability no vendor sells, and reuse existing Accio datasets, skills, and connectors instead of rebuilding them per application.
- Take each application through the same evaluation, controls, and release evidence as the shared products, and own its support and lifecycle once live or hand it to a clear owner.
Build in evaluation and control, and work with the team- Build evaluation and regression suites for models, prompts, agents, and platform changes, meet the agreed thresholds before release, and instrument what you build for quality, safety, reliability, latency, drift, usage, and cost.
- Implement the controls defined by AI Governance Implementation and AI Security as code and secure defaults, and produce release evidence proportionate to the risk of the use case, generated by the platform rather than assembled after the event.
- Work alongside Percepta engineers on Libros, PRISM, and the wider estate, holding joint work to our engineering standards and raising maintainability problems while they are still cheap to fix.
- Mentor AI Engineers through pairing, design discussion, and code review, and help establish the team's engineering standards, agentic SDLC, and definition of done.
- Work with AI Architecture to turn reference patterns into implementations teams use, partner with Forward Deployed Engineering so proven solutions become supported shared capability, and feed adoption data from the AI Enablement Partners into the backlog.
Must have skills- At least six 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, covering prompt and context engineering, agent development, retrieval, tool use, evaluation, and deployment at scale.
- Strong Python and SQL, with the judgement to write code others can maintain and extend.
- Hands-on experience building agentic capability such as MCP servers, tools, skills, or connectors for other systems and agents to consume.
- Hands-on experience with a major cloud, ideally Azure, including containers or serverless compute, infrastructure as code, CI/CD, identity, RBAC, and secret management.
- Experience owning services in production - monitoring, incident investigation, upgrades, lifecycle management - and establishing evaluation and observability for AI systems.
- Good judgement in a regulated environment, translating security, privacy, risk, and audit requirements into working technical controls.
- Experience mentoring engineers and leading technical work without relying on reporting authority, and clear communication with engineers, control functions, and business stakeholders.
Nice to have skills- Asset management or financial services domain knowledge, particularly the investment process, distribution, or front-office workflows.
- Azure AI Foundry, Azure OpenAI, Anthropic Claude, model gateways, inference routing, or agent orchestration platforms.
- Snowflake, Microsoft Fabric / OneLake, or comparable governed enterprise data platforms.
- TypeScript or a second production language, and front-end experience for user-facing AI applications.
- Experience building internal developer platforms or golden-path patterns used by other engineering teams, or taking partner-built software into internal ownership.
What to expect when you join our firm
- Hybrid working and reasonable accommodations
- Generous Holiday policies
- Excellent Health and Wellbeing benefits including corporate membership to Wellhub
- Paid volunteer time to step away from your desk and into the community
- Support to grow through professional development courses, tuition/qualification reimbursement and more
- Maternal/paternal leave benefits and family services
- Unique employee events and programs including a 14er challenge
- Complimentary beverages, snacks and all employee Happy Hours
Supervisory responsibilitiesNo. This is a senior individual-contributor role. The Senior Applied AI Engineer leads technical delivery, sets implementation standards, and mentors AI Engineers, but does not line-manage. The role may act as technical lead for a product or platform workstream.
Compensation information
The base salary range for this position is $130,000 - $220,000. This range is estimated for this role. Actual pay may be different. This position will be open through [add date]
