Platform Architect

Responsibilities:

- Build the AI platform architecture: the reference design for how LLM capabilities are accessed, governed, monitored, and extended across the organisation.

- Own every architectural decision for the AI platform, and document, justify, and stand behind each one.

- Evaluate and select new models, tools, and frameworks as the market moves.

- Deliver the platform as a service, so teams can move their agents into governed cloud infrastructure.

- Govern Anthropic (Claude API, Claude Code, Anthropic Enterprise), OpenAI, and future providers, covering API access controls, credential management, model version policy, cost allocation, and acceptable use standards.

- Own technical relationships with Anthropic and OpenAI at the enterprise level.

- Build cloud infrastructure for AI workloads on AWS and GCP: Bedrock, SageMaker, Vertex AI, Cloud Functions, and supporting services (S3, Cloud Storage, IAM, and monitoring on both platforms).

- Deploy and securely configure n8n, and build automation workflows connecting LLMs to Jira, Confluence, Slack, Google Workspace, HR systems, and proprietary data sources.

- Design and secure MCP servers and gateways connecting AI agents to internal services.

- Build Python integrations for AI-powered internal tooling: RAG pipelines, agent frameworks, prompt libraries, and internal AI assistants.

- Deliver hands-on: you write the code, build the workflows, and deploy the tooling. This isn't an advisory role.

- Build data privacy and security architecture for LLM usage in a regulated environment, including prompt classification, output controls, and audit trails aligned with FCA, CySEC, ASIC, and DORA.

- Build observability across the full stack: token consumption, cost per team, latency, error rates, and policy violations across cloud platforms and direct API providers.

- Build an AI adoption framework of standards, patterns, and onboarding materials that let teams across Technology use AI safely and consistently.

- Run internal enablement through workshops, documentation, and direct support for teams moving from ad-hoc AI use to governed, repeatable processes.

- Report AI platform status, risks, and roadmap to Technology leadership

Requirements:

- 8+ years in engineering, with at least 2 years building LLM-based systems in production, not prototypes or POCs.

- Comfort operating as the first person in a function: you set your own priorities, unblock yourself, and deliver without a team around you.

- Production-grade Python: tooling that others depend on, not scripts you hand off.

- Deep, hands-on experience with the Anthropic and/or OpenAI APIs: function calling, tool use, streaming, RAG, prompt management, token optimisation, and model versioning.

- Hands-on experience with agentic platforms, MCP servers, and MCP gateways, including designing flows between AI agents and internal services.

- n8n: deploying, securely configuring, and maintaining multi-step automation workflows at organisation scale.

- Solid AWS for AI workloads: Bedrock, SageMaker, Lambda, S3, IAM, and CloudWatch, deployed and operated in production.

- Solid GCP for AI workloads: Vertex AI, Cloud Functions, Cloud Storage, IAM, and Cloud Monitoring.

- Hands-on DevOps capability to deploy and run what you build.

- Architectural thinking: end-to-end system design, clear trade-off documentation, and the ability to defend decisions to senior stakeholders.

- Security engineering for what you build: secrets and credential management, key rotation, and least-privilege IAM across cloud and SaaS.

- Data protection and auditability by design: classifying what data can reach which model and provider, controlling and logging prompts and outputs, and building audit trails that hold up to a regulator.

- Fluency in the LLM threat surface, including prompt injection, data exfiltration through model outputs, insecure tool and agent use, and model supply-chain risk, with working knowledge of a recognised framework such as the OWASP LLM Top 10 or NIST AI RMF.

Nice to have:

- Fintech, brokerage, or financial services background.

- DORA or ISO 27001:2022 knowledge as it applies to AI system risk.

- Familiarity with the EU AI Act and other AI regulatory requirements, as well as AI-specific risk and governance standards such as NIST AI RMF or ISO/IEC 42001.

- LLM orchestration frameworks: LangChain, LlamaIndex, or similar.

- Hands-on experience with LLM guardrail and DLP tooling: Bedrock Guardrails, Vertex AI safety filters, or prompt-firewall and PII-redaction tools.

- Experience building internal developer platforms or AI centres of excellence.

- AWS Solutions Architect Professional, Google Professional ML Engineer, or equivalent.