AI Engineer

AI Engineer - Risk Testing AI Assistant - Contract - London/Hybrid - £650pd via Umbrella 4 days per week onsite £650per day via Umbrella Build and deliver an AI agent that integrates with QA environment to automate risk testing workflows.

The initial focus is Market Risk (Vanilla VAR and Stress Testing), with the goal of proving the system before scaling to other risk analysis types.

This work forms part of broader agentic SDLC programme across Risk.

Required Technical Skills

• Agentic AI Development (Primary)
• Python: Strong production-level Python. This is the primary language for agent development.
• LangChain/LangGraph: Hands-on experience building multi-agent workflows with state management, human-in-the-loop patterns, and tool orchestration. This is standard agent framework.
• Model Context Protocol (MCP): Experience building MCP Servers and integrating MCP tools into agent workflows. Is an early adopter with production MCP integrations for Control-M, Confluence, Jira, and other systems.
• LLM Integration: Practical experience working with OpenAI GPT models (GPT-4.1, GPT-5.x), prompt engineering, and model evaluation. Understanding of token economics, cost optimisation, and model selection trade-offs.
• RAG Pipelines: Experience with retrieval-augmented generation using vector search and hybrid retrieval (BM25 + vector).
• Azure Cloud Platform
• Azure AI Foundry: Model deployment, prompt management, and evaluation workflows.
• Azure OpenAI Service: Deploying and managing LLM endpoints.
• Azure AI Search: Building and maintaining search indexes for RAG.
• QA/Test Automation
• Test automation frameworks: Experience automating QA workflows, test data generation, and test case creation.
• BDD (Behaviour-Driven Development): Familiarity with BDD frameworks. Uses Vividus Copilot for BDD test automation.
• CI/CD: GitHub Actions for pipeline automation and integration testing.
• Environment provisioning: Automating setup and configuration of test environments.
• Development Practices
• Git/GitHub: Version control, branching strategies, pull requests, and code review.
• VS Code: Primary IDE. Experience with GitHub Copilot integration and custom agents/extensions is a strong plus.
• API development: FastAPI or Flask for building Back End services.
• Observability: Experience with Azure Monitor, Application Insights, or LangSmith for LLM application monitoring.

Required Domain Knowledge

• Capital Markets/Risk (Preferred, not mandatory)
• Understanding of Market Risk concepts: Value at Risk (VAR), Stress Testing, risk scenario modelling.
• Familiarity with risk systems (eg Murex, Raptor) or similar trading/risk platforms.
• Awareness of QA processes within financial services, including regulatory testing requirements.

Security and Governance Awareness

• Operates with banking-grade controls.
• Engineers must be comfortable working within: AI Governance (AIG) approval processes and Architecture Review Board (ARB) gates.
• Secure-by-Design evidence requirements and CVE scanning.
• Strict data classification, DLP, and sensitivity labelling (Microsoft Purview).
• Private networking (VNet, Private Endpoints, ExpressRoute).
• No public Internet access for production workloads.
• Audit trail requirements with immutable logging.
• DPIA (Data Protection Impact Assessment) processes.

Desirable Skills (Nice to Have)

• Experience with Semantic Kernel (C) as an alternative orchestration framework.
• Familiarity with Control-M batch scheduling and job orchestration.
• Experience with Jira and Confluence integration (API-level).
• Exposure to other risk types: Credit Risk, Counterparty Risk.
• Experience working within regulated financial services environments.
• Understanding of agentic SDLC patterns: autonomous decision levels, adversarial review loops, agent specialisation (Planner, Implementer, Reviewer).