AI Engineer

We are building a new R&D capability focused on developing data-driven and AI-enabled solutions for legal services and the wider business of law. The AI Engineer will build end-to-end AI capabilities for R&D products: agentic workflows, retrieval and context, agent and evaluation harnesses, and the user-facing features around them. Working with the Head of R&D, you will translate agreed product and technical designs into secure, observable and maintainable products.

This is a hands-on full-stack AI engineering role, with applied AI product delivery at its core. You will build AI workflows, tools and integrations, evaluation and observability capabilities, together with the APIs, services and user interfaces needed to ship them reliably. This is not a research-only role. You will apply agreed enterprise architecture, security and deployment patterns, while taking practical responsibility for code quality, testing, monitoring and troubleshooting with product, data, technology and security teams.

You will join a small, hands-on multidisciplinary team. Each team member will work collaboratively across discovery, prototyping, engineering, productionisation and continuous improvement.

• Build complete AI-enabled product features across user interface, backend, AI orchestration, retrieval and approved enterprise integrations.
• Design and build agentic AI workflows using structured outputs, tool calling, workflow orchestration, document processing, human-review steps and appropriate controls.
• Build reusable agent harnesses: practical runtime patterns around tools, state, context, approvals, traces and test fixtures that make agents consistent, inspectable and safe to use.
• Build grounded retrieval and context patterns using approved data products, search indexes, document repositories and permission-aware sources; provide evidence and citations where appropriate.
• Design and operate evaluation harnesses for AI workflows, including curated test sets, representative scenarios, automated and human grading, regression tests, trace review and user-feedback loops.
• Manage prompts, model configuration, tool schemas and routing as tested product assets, including fallbacks and cost/latency trade-offs.
• Implement AI observability through traces, logs, metrics and evaluation results, so quality, reliability, failure modes, latency and cost are visible and can be improved.
• Build clear APIs and integrate AI products with approved data, document, search and business systems.
• Build usable interfaces that help users understand AI output, inspect supporting evidence, provide feedback and complete review or approval steps.
• Write maintainable Python and Type Script code; use Docker, Git, automated testing and CI/CD to make development, testing and deployment repeatable.
• Apply agreed security, authentication, data-handling, deployment and release patterns, working with the relevant firm technology teams where needed

• End-to-end agentic AI features and reusable harnesses for Radar and other named R&D products.
• A practical evaluation and test capability that lets the team compare prompts, models, tools and retrieval approaches before and after release.
• Grounded AI workflows that combine firm data, documents, market or regulatory information and user context into useful, attributable outputs.
• Interfaces that make AI output, confidence, evidence, controls and next actions clear to lawyers and business users.

• R&D products use AI workflows that are grounded in trusted context, observable in operation and understandable to users.
• AI quality is measured rather than assumed; evaluation results, traces and feedback lead to demonstrable improvements in usefulness, reliability, latency and cost.
• The engineer can independently take a well-scoped AI workflow from problem definition to controlled deployment.
• At least one AI component, harness or evaluation capability is reused by a second R&D product or workflow.
• Promising prototypes have a repeatable path to controlled release, monitoring and improvement.
• At least three years' professional experience delivering production AI products, with demonstrable recent experience building AI-enabled products or workflows in a SaaS environment. Experience in legal technology or a law firm is not required, but would be advantageous.
• Strong Python, plus practical Type Script or JavaScript. Able to build APIs and services using FastAPI, Flask or comparable technologies, and usable interfaces…