Location
Paternoster Square, London

Hours
Full Time

Salary
Negotiable

About the Role
The Corporate Engineering AI (CE AI) team at London Stock Exchange Group (LSEG) is the central enablement and platform delivery function for LSEG’s internal agentic AI ecosystem. The team’s mission is to scale safe, high-quality AI capabilities across the enterprise by providing shared platforms, patterns, governance, and delivery support. CE AI owns and operates core AI platforms including LSEG AI Assist, the Question Answering Service (QAS), and the Internal MCP Gateway.

Rather than delivering individual business use cases end-to-end, the team enables product engineering groups across LSEG to expose knowledge, data, and actions to AI agents in a consistent, governed, and repeatable way. The team operates a Central MCP Delivery model: building critical MCP tools and services “for” product teams where required, while simultaneously defining standards, patterns, and platform capabilities that allow teams to progressively move towards self-service contribution.

The LSEG AI Assist / Internal MCP Programme of Work delivers an LSEG-owned, production-grade agentic AI platform with MCP as its extensibility layer. This includes building and operating LSEG AI Assist, an in-house agentic experience capable of reasoning, planning, and tool-calling; operating QAS, the enterprise RAG and search layer used to ground agent responses in approved data sources; delivering a production Internal MCP Gateway providing discovery, security, policy enforcement, observability, and lifecycle management for MCP tools and Skills; designing and building MCP servers and Skills that expose internal and vendor systems safely to agents; and establishing evaluation, quality control, and governance mechanisms so MCP tools and Skills can be promoted through Permit to Build (PTB) and Permit to Operate (PTO) and operated with confidence at scale.

The programme follows a “build for” model today, with a strong emphasis on defining the future product and platform experience, patterns, and contribution pathways that will enable federated scale over time.

Experience
- Strong Python experience for test harnesses and automation
- Experience with LLM and RAG evaluation frameworks or custom evaluation pipelines
- Test automation expertise covering unit, integration, and regression testing
- Understanding of agentic system risks and failure modes
- Ability to assess solutions against governance, security, and audit expectations
- Experience working in regulated or highly governed engineering environments

About you
You are a Senior Associate level professional with a strong commitment to quality and safety in AI systems. You have a keen eye for detail and a proactive approach to defining and implementing evaluation frameworks. You thrive in collaborative environments, partnering closely with ML Engineers and other stakeholders to embed testability and governance into AI platforms. You are motivated by innovation and continuous improvement, and you are comfortable working within regulated or highly governed settings.

Qualifications
Relevant experience in machine learning quality assurance, software testing, or related fields is essential. Formal qualifications in computer science, engineering, or a related discipline are advantageous but not mandatory. A strong understanding of AI agentic systems and their associated risks is critical.

Responsibilities
- Define and implement evaluation frameworks for MCP tools and Skills covering correctness, safety, and regression impact
- Build and maintain automated test pipelines for agentic behaviours, including tool invocation and multi-step workflows
- Evaluate and mitigate agentic failure modes such as hallucination, tool misuse, invalid inputs, and latency amplification
- Produce testing evidence required for Permit to Build (PTB) and Permit to Operate (PTO)
- Partner with ML Engineers to embed testability and evaluation hooks into MCP servers and Skills
- Help define the long-term quality and governance model for federated MCP contributions across LSEG