AI & Systems Architect

HM Revenue & Customs; United Kingdom; Vor Ort; GBP 90.000 - 140.000; Vollzeit

Zusammenfassung

HM Revenue & Customs seeks a senior architect to lead the architecture and engineering of an AI-assisted decision-systems prototype. You will design agentic workflows, evaluation framework and control boundaries that connect authoritative source material to controlled representations, a deterministic language, standard data, and downstream implementations.

Lead architecture across AI, data, and assurance, delivering observable artefacts, tool contracts, and regression tests while ensuring

Qualifikationen

• Own the end-to-end technical architecture and engineering roadmap for the prototype.
• Design multi-agent workflows for extraction, decomposition, transformation, validation, comparison, and explanation of complex rule sets.
• Define where generative AI is appropriate and where deterministic code, schemas, constraints, and human approval are required.
• Engineer structured outputs, tool/API contracts, evaluation harnesses, golden datasets, error taxonomies, and regression tests.
• Work closely with the language/data architect so agent outputs conform to the DSL grammar, standard schema, and graph model.
• Create observable background across source evidence, agent transformations, human decisions, and generated artefacts.
• Provide technical leadership across AI, data, SAP, and assurance specialists.
• Deep experience designing and delivering GenAI/LLM systems beyond proof-of-concept prompting.
• Hands-on experience with agent orchestration, tool use, structured outputs, retrieval, and evaluation.
• Strong API/software-engineering skills, including automated testing.
• Experience designing AI quality measures, failure handling, human-in-the-loop controls, and auditability.
• Ability to lead architecture across AI, data models, line graphs, and enterprise systems.
• Knowledge representation, formal methods, or rules engines.
• Compiler/DSL awareness sufficient to collaborate on constrained generation.
• Knowledge graphs or graph databases such as Neo4j/RDF-based technology.
• Experience delivering AI in regulated, financial, government or other high-assurance environments.

Aufgaben

• Own the end-to-end technical architecture and engineering roadmap for the prototype.
• Design multi-agent workflows for extraction, decomposition, transformation, validation, comparison, and explanation of complex rule sets.
• Define where generative AI is appropriate and where deterministic code, schemas, and human approval are required.
• Engineer structured outputs, tool/API contracts, evaluation harnesses, golden datasets, error taxonomies, and regression tests.
• Work closely with the language/data architect so agent outputs conform to the DSL grammar, standard schema, and graph model.
• Create observable background across source evidence, agent transformations, human decisions, and generated artefacts.
• Provide technical leadership across AI, data, SAP, and assurance specialists.
• Deep experience delivering GenAI/LLM systems beyond proof-of-concept prompting.
• Hands-on experience with agent orchestration, tool use, structured outputs, retrieval, and evaluation.
• Strong API/software-engineering skills, including automated testing.
• Experience designing AI quality measures, failure handling, human-in-the-loop controls, and auditability.
• Ability to lead architecture across AI, data models, line graphs, and enterprise systems.
• Knowledge representation, formal methods, or rules engines.
• Compiler/DSL awareness sufficient to collaborate on constrained generation.
• Knowledge graphs or graph databases such as Neo4j/RDF-based technology.
• Experience delivering AI in regulated, financial, government or other high-assurance environments.

Jobbeschreibung

Lead the architecture and engineering of an AI-assisted decision-systems prototype. You will design the agentic workflow, evaluation framework and control boundaries that connect authoritative source material to controlled representations, a deterministic language, standard data, and downstream implementations.

What you do

• Own the end-to-end technical architecture and engineering roadmap for the prototype.
• Design multi-agent workflows for extraction, decomposition, transformation, validation, comparison, and explanation of complex rule sets.
• Define where generative AI is appropriate and where deterministic code, schemas, constraints, and human approval are required.
• Engineer structured outputs, tool/API contracts, evaluation harnesses, golden datasets, error taxonomies, and regression tests.
• Work closely with the language/data architect so agent outputs conform to the DSL grammar, standard schema, and graph model.
• Create observable background across source evidence, agent transformations, human decisions, and generated artefacts.
• Provide technical leadership across AI, data, SAP, and assurance specialists.
• Deep experience designing and delivering GenAI/LLM systems beyond proof-of-concept prompting.
• Hands-on experience with agent orchestration, tool use, structured outputs, retrieval, and evaluation.
• Strong API/software-engineering skills, including automated testing.
• Experience designing AI quality measures, failure handling, human-in-the-loop controls, and auditability.
• Ability to lead architecture across AI, data models, line graphs, and enterprise systems.
• Knowledge representation, formal methods, or rules engines.
• Compiler/DSL awareness sufficient to collaborate on constrained generation.
• Knowledge graphs or graph databases such as Neo4j/RDF-based technology.
• Experience delivering AI in regulated, financial, government or other high-assurance environments.