Lead AI Solutions Architect – 6 mths - £750-£850 – hybrid)
Our public sector client is seeking an experienced Lead AI Solutions Architect for 6 months on a flexible hybrid basis based from either their London, Manchester or Bristol office.
The team is responsible for innovation and digital design of AI tools for policy making in the Civil Service.
Experience working in or with the UK government is essential.
Familiarity with government AI guidance and responsible AI principles is required.
The Lead AI Solutions Architect will provide senior technical advice to the team on the practical application of AI, knowledge technologies and digital tools within policymaking, translate the teams priority hypotheses into credible technical options, experiments and pilot designs that can be tested in live OGD policy environments. Develop technical options for testing hypotheses.
Essential experience:
An overarching requirement to be someone who can combine AI architecture with delivery expertise.
Strong professional experience designing and delivering AI-enabled products from experiments to prototypes, testing and evaluation.
Demonstrable knowledge of generative AI and agentic systems.
Experience with knowledge architecture and information retrieval – including metadata, semantic search and knowledge graphs for structuring, connecting, organising and retrieving knowledge.
Strong solution design and architecture skills.
Experience integrating AI with existing data, document and workflow environments.
Strong understanding of AI technology, its limitations, and risks – including hallucination, bias, data quality, model variability, information security, privacy, explainability, and over-reliance on automated outputs.
Experience working in multidisciplinary teams and presenting to non-technical audiences.
Experience designing for reuse and scalability – including developing reference architecture, reusable technical components and implementation guidance.
Familiarity with government AI guidance and responsible AI principles is required
