Axiologik are looking to bring in a Contract SC Architect with experience of working on AI projects. This is a high value piece of work so if you love seeing the impact of your work come to life, read below and apply.
£750 DR
Inside IR35
12 month contract - extensions likely
Remote with sporadic travel to London (client offices)
MUST HAVE ACTIVE SC CLEARANCE
The role will predominantly be to own the technical architecture for a new AI product, develop machine learning models supporting stakeholders and operationalise that capability.
The architect will own the end-to-end design across model training, inference and integration into the existing estate and to set and hold the target architecture through build and delivery. This role will also act as the technical authority for both supplier and internal engineering teams.
Architecture
- Proven technical/solution architect on AI-focussed technology projects within larger, strategic programmes, owning designs from concept through to live service
- Able to produce and govern the full artefact set namely HLDs, LLDs, architecture decision records, options papers with costed trade-offs, and target/transition state architectures.
- Experience of taking designs through formal design authority and technical assurance boards and holding the design under challenge
- Broad understanding of solution architecture, architectural delivery, process, modelling and notation within enterprise-level contexts (TOGAF, ArchiMate, C4, UML, BPMN)
- Non-functional design at scale considering throughput, latency budgets, availability, resilience and disaster recovery across a distributed multi-site estate
- Security architecture experience at OFFICIAL-SENSITIVE, applying Secure by Design and NCSC guidance
AI and ML
- Deep, hands-on understanding of AI/ML system design rather than conceptual familiarity including model serving topologies, inference scaling, and where the accuracy/latency/cost trade-offs sit
- Experience in hosting and tuning AI models to achieve scalable AI inference for users
- Working knowledge of computer vision model architectures (CNNs, vision transformers) and detection/segmentation approaches, sufficient to challenge and direct data science teams
- Model optimisation for constrained and edge environments for example quantisation, pruning, distillation, ONNX/TensorRT conversion, accelerator selection and sizing
- Experience with AI/ML platforms and processes eg OpenVINO or similar toolkit, MLOps training pipeline or SageMaker
- Architecture of the model lifecycle including design of data and feature pipelines, model registry and versioning, CI/CD for models, drift detection, monitoring and retraining triggers
- Ability to define an evaluation strategy in operational terms condsidering precision/recall trade-offs, threshold setting, and the false-positive burden placed on stakeholders
- Design of human-in-the-loop patterns where model output informs, rather than replaces, officer decision-making
Platform and Data
- In-depth AWS design experience, particularly S3, serverless computing (Lambda) and EventBridge/SNS/SQS
- Edge and hybrid architecture including inference at or near port sites under constrained connectivity, with cloud-based training and central model distribution
- Data architecture for high-volume imagery providing design for storage tiering, retention, lineage and provenance
- Experience of data governance, privacy policies and AI assurance methods (DPIAs, model documentation, transparency and bias/fairness assessment)
Delivery and Stakeholder
- Experience working with both supplier and DevOps teams in complex delivery environments to ensure that deployed solutions meet stakeholder expectations
- Able to act as technical assurance over supplier designs, including build-vs-buy and vendor lock-in assessment
- Strong senior stakeholder management experience
- Excellent communication skills and the ability to translate architecture into understandable concepts for senior business stakeholders
- Comfortable setting architectural direction and standards for engineering teams, and coaching those teams to work within them