Saas & AI Solution Architect

Solution Architect Contract, roughly 7 months, UK based, hybrid. Reports to the CTO. A senior, hands-on architecture role, not advisory. Role overview The Solution Architect owns the technical shape of the whole platform and makes the hardest requirements real: strict multi-tenant isolation, post-based accountability that stays correct across time, an operational-tree data model that doubles as the structured substrate the AI reads, and a durable workflow layer. You set the standards the entire team builds to, make the load-bearing decisions, and stay close enough to the code to prove your designs work. You will be designing developing a Saas platform with AI and automation capabilities. Key responsibilities on this project • Own the end to end architecture: the multi-tenant PostgreSQL data model, service boundaries, the workflow layer on Temporal, the AI retrieval and grounding design, and the AWS topology. • Design database-level tenant isolation using PostgreSQL row-level security, and prove it cannot be bypassed. • Design the post-based, effective-dated (bitemporal) accountability model, so the platform can answer who held which post on any past date without rewriting history. • Design the per-company operational tree as a self-referencing hierarchy with recursive queries, and make it the structured graph the AI agents traverse. • Set and enforce the no-raw-SQL standard (ORM and parameterised access only) and the curated, read-only tool surface the AI is allowed to use. • Own the threat model, the security baseline and the UK data-residency design (UK region, UK GDPR) with the independent security specialist. • Run the weekly architecture review with the client COO, and record the decisions and their rationale. • Define the non-functional targets (performance, availability, disaster recovery) and hold the build to them. Essential skills and experience • 10+ years in software engineering, with 4+ years in a hands-on architecture role. You have owned the architecture of a production system, not just advised on one. • Deep PostgreSQL. Data modelling at scale, row-level security, recursive CTEs, indexing and query performance, and temporal or effective-dated data patterns. • Multi-tenant SaaS. You have designed and shipped a genuinely multi-tenant product with hard tenant isolation, and you understand the trade-offs between the isolation models. • AWS. ECS or EKS, RDS, ElastiCache, S3, CloudFront, SES, IAM, infrastructure as code, and cost-aware design. Experience in developing with •Temporal creating automated workflows • Modern application stack. TypeScript with NestJS or Node, and Python with FastAPI; React and Next.js on the front end; REST API and event-driven design. • Workflow and durable execution. Hands-on with a durable-workflow engine (Temporal strongly preferred) or comparable orchestration, with a clear view of when to adopt versus build. • AI-aware architecture. You understand retrieval-augmented generation, vector search (pgvector), grounding, evaluation and auditing well enough to design the data and tool surfaces the agents depend on. • Security and compliance by design. Threat modelling, secure-by-design, and UK GDPR or equivalent data-protection design. Desirable • Experience replacing or migrating a large legacy system (100+ tables) onto a modern platform. • Public-sector, regulated or government-adjacent delivery. • Prior use of Auth0 or AWS Cognito at scale. • Exposure to multiple LLM providers (Claude, GPT-4o) through a provider-agnostic abstraction.