Senior Principal Architect, CTO Office

The role

This is a hands-on architect role that grows into a player-coach role. You will start as the senior technical individual contributor in the CTO Office and then hire and lead a small team of engineers built in your image.

The unit takes on whatever matters most at the time. One month that is a production-grade agentic system with a frontier model provider, the next it is an architecture review of a platform under consideration, an internal AI tool the whole company adopts, or a technical escalation from a strategic customer. There is no standing backlog to hide behind. You go in, learn the domain fast, deliver something real in weeks, and hand it over or harden it.

We are explicit about the calibre we are hiring for. The team exists to produce output that would normally take a squad a quarter, in a fraction of the time, at production quality. If that sounds exhausting rather than energising, this is the wrong role.

What you will do

• Own architecture and delivery on the CTO Office's highest-priority technical projects, end to end
• Write and review code daily; the role never becomes slideware
• Design and ship agentic and LLM-based systems to production standard: retrieval at scale, evals, guardrails, security and identity, monitoring and auditability
• Drop into unfamiliar codebases, stacks, and business domains and become dangerous within days
• Hire, coach, and set the technical bar for a small team of applied AI engineers
• Brief executives and senior customer stakeholders on what you built and what it means commercially

Qualifications

What we are looking for

• 15+ years building software, with architecture ownership of systems running in production at enterprise scale
• Still keyboard-credible: you review PRs, debug agents, and ship. We will verify this in the interview process
• Deep, current applied AI capability: production agentic systems, RAG and Graph RAG, MCP, agent runtimes, and fluency across model providers rather than loyalty to one
• Eval fluency. You can articulate and build an evaluation strategy for an LLM system. Candidates who cannot are not a fit, regardless of seniority
• Evidence of operating AI systems at enterprise scale, not a portfolio of PoCs
• Range and speed: a track record of delivering across different domains, stacks, and problem shapes under time pressure
• Experience hiring engineers and raising a team's technical standard
• Clear written and spoken communication at executive level

Technical depth expected

We hire for depth per area, not a keyword count. The named tools are examples; strong equivalents are fine. Expect to be tested on the areas below, not asked whether you have heard of them.

• Languages: expert-level Python and TypeScript for AI and product work, and the ability to read and contribute to the large enterprise codebases you will drop into (C#/.NET, Java)
• Model providers: hands-on production work with at least 2 of Anthropic, OpenAI, Google, and open-weight models (direct APIs or via Azure AI Foundry, Bedrock, Vertex); tool use and function calling, structured outputs, context and token management, prompt architecture
• Agentic systems: designing, hosting, and operating autonomous and semi-autonomous agents in production; MCP servers and clients; orchestration frameworks such as LangGraph, Pydantic AI, or provider-native SDKs; sandboxing, permissioning, and human-in-the-loop design
• Retrieval and knowledge: RAG and Graph RAG at enterprise scale; vector and hybrid search (pgvector, OpenSearch/Elasticsearch, or dedicated stores) including chunking strategy, reranking, and distributed retrieval; knowledge graphs and entity resolution; Text-to-SQL over complex enterprise schemas
• Evals and quality: golden datasets, regression suites, LLM-as-judge patterns, online monitoring of model behaviour; tooling such as RAGAS, promptfoo, Langfuse or LangSmith, or eval harnesses you built yourself
• Security, identity, and governance: OAuth2/OIDC, Entra ID and service principals, secrets management; guardrails and content safety; auditability and traceability of agent decisions
• Cloud and platform: production Kubernetes and containerisation, infrastructure as code (Terraform or Bicep), CI/CD; deep in at least 1 hyperscaler (Azure preferred given our estate) with working literacy in the others
• Observability and operations: OpenTelemetry or equivalent distributed tracing applied to LLM systems, cost and latency engineering, capacity planning for tive maintenance
• Industrial, asset-intensive, or enterprise software domain exposure (EAM, ERP, FSM)
• Working knowledge of AI governance frameworks such as the EU AI Act and NIST AI RMF
• Prior 0-to-1 work: standing up a capability, product, or team from nothing

How we assess

A walkthrough of systems you have personally architected and shipped, a technical deep dive including your eval and production-hardening approach, and a working session on a live problem. References will be asked specifically about hands-on contribution, not leadership style.

Practicalities

• Location and working model: Virtual/Hybrid – location dependent
• Compensation: Upon discussion
• Some international travel

Additional Information

We embrace flexibility and hybrid work opportunities to support diverse needs and lifestyles, while also valuing inclusive workplace experiences. By fostering a sense of community, we drive innovation, strengthen connections, and nurture belonging. Our commitment ensures you can work in a way that suits you best, while also engaging with colleagues to share ideas and build meaningful relationships.