Senior AI Solutions Architect

𝗦𝗲𝗻𝗶𝗼𝗿 𝗔𝗜 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝘀 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁 (𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜, .𝗡𝗘𝗧, 𝗔𝘇𝘂𝗿𝗲)

𝗟𝗼𝗰𝗮𝘁𝗶𝗼𝗻: Remote

𝗣𝗼𝘀𝗶𝘁𝗶𝗼𝗻 𝘁𝘆𝗽𝗲: Contract

𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲: 8-12 years, of which at least 2 in production LLM or agent-based systems.

𝗔𝗯𝗼𝘂𝘁 𝘁𝗵𝗲 𝗿𝗼𝗹𝗲

This is a hands-on architecture role on an AI-led enterprise application. You will own the solution and technical architecture end to end - the web application, the services around it, and the AI tier that sits between a user's question and the systems that can answer it.

𝗦𝘁𝗮𝗰𝗸: .NET / ASP.NET Core, React or Angular, Azure, Entra ID, MCP and agent orchestration.

𝗠𝗮𝗻𝗱𝗮𝘁𝗼𝗿𝘆 𝘀𝗸𝗶𝗹𝗹𝘀 - 𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹

𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲

• Has led architecture engagements in full-stack web application development, not only advised on them.
• Has worked as an architect alongside tech leads, multiple engineering teams and multiple developers.
• Full-stack delivery on the .NET stack - ASP.NET Core, Web API, C#, with .NET Framework or MVC legacy exposure useful for modernisation work. A strong Java (Spring Boot) or Node/TypeScript background with .NET willingness will be considered.
• Modern front end: React or Angular with TypeScript, component architecture, and streaming or incremental UI for long-running AI responses.
• RESTful and event-driven service design, API-led connectivity, versioning and backward compatibility across consumers.
• Data structures, design patterns, and the judgement to know when a pattern is costing more than it returns.
• High-level and low-level design as written artifacts - sequence, component and deployment views, and a non-functional section that survives a security review.

𝗔𝗜 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴: 𝗮𝗴𝗲𝗻𝘁𝘀, 𝗠𝗖𝗣 𝗮𝗻𝗱 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻

• Production experience with LLM applications: prompt and context design, structured output, function and tool calling, streaming, and failure handling when the model returns something unusable.
• Agentic patterns - planner/executor, tool-using agents, multi-agent hand-off, human-in-the-loop checkpoints - and a view on where each is warranted.
• Model Context Protocol: building and operating MCP servers, designing tool and resource schemas, and passing caller identity through to the underlying system rather than running everything as a service account.
• Custom orchestration logic in code. Familiarity with Semantic Kernel, LangGraph, LangChain or AutoGen is useful, but we expect you to know what the framework is doing and to be comfortable replacing it where it does not fit.
• Retrieval design: chunking and indexing strategy, hybrid vector and keyword search, reranking, citation, and entitlement filtering applied at retrieval time, not after.
• Guardrails and safety controls appropriate to an enterprise portal: PII handling, prompt injection resistance for tool-using agents, output validation, and full audit logging.
• Evaluation as engineering, not sampling by eye - golden question sets, regression gates in CI, tracing and observability, and cost and latency measured per interaction.

𝗜𝗱𝗲𝗻𝘁𝗶𝘁𝘆, 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗮𝗻𝗱 𝗮𝗰𝗰𝗲𝘀𝘀

• Enterprise authentication and authorisation with Active Directory or Microsoft Entra ID - SSO, OIDC and SAML, on-behalf-of flows, service principals, and managed identities.
• Group-based entitlement design carried through the full path: portal, services, agents, tools and the semantic model, including row-level and object-level security.
• Audit, traceability and data-residency considerations for AI features, and the ability to hold that conversation with a customer's security team without escalating everything.

𝗖𝗹𝗼𝘂𝗱, 𝗰𝗼𝗻𝘁𝗮𝗶𝗻𝗲𝗿𝘀 𝗮𝗻𝗱 𝗱𝗲𝗹𝗶𝘃𝗲𝗿𝘆

• Built and deployed applications on a public or private cloud - Azure preferred given the AD and data platform context, with AWS or GCP acceptable.
• Docker and Kubernetes in anger: workload sizing, secrets, configuration, scaling behaviour for GPU or token-bound workloads, and what to do when a pod restarts mid-conversation.
• CI/CD with Azure DevOps or GitHub Actions, infrastructure as code, and prompt and model versioning treated with the same discipline as application code.
• An active role in development, code review, continuous integration and deployment. This is not a diagram-only role.

𝗠𝗮𝗻𝗱𝗮𝘁𝗼𝗿𝘆 𝘀𝗸𝗶𝗹𝗹𝘀 - 𝗻𝗼𝗻-𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹

• Has worked directly with customer and business stakeholders, business analysts and project managers to shape and deliver business use cases.
• Comfortable with abstract problem statements in ambiguous customer environments, and able to work independently and give technical leadership to the team.
• Can explain an architectural trade-off to a non-technical sponsor and a security reviewer in the same meeting, and take a position rather than list options.
• Strong first-principles thinking. Depth over breadth of tool names.
• Strong written and spoken English; strong analytical, co-ordination and organisational skills.

𝗣𝗿𝗲𝗳𝗲𝗿𝗿𝗲𝗱 𝘀𝗸𝗶𝗹𝗹𝘀

• Enterprise application integration and distributed systems architecture - microservices, event-based integration, ESB or iPaaS exposure.
• Knowledge graphs, ontologies or metadata catalogues supporting AI retrieval.
• Legacy modernisation and pre-sales or discovery experience - sizing, phasing and naming uncertainty in front of a customer.
• Contributions to internal capability building: reusable accelerators, reference architectures or practice development around AI and integration.