Applied AI Engineer in London

The Department

Our Technology team continues to go from strength to strength, playing a vital role in firm-wide innovation and strategy.

Far from being a support function, Technology is integral to the firm's growth agenda, backed by strong leadership engagement and sustained investment.

The Team

The AI and Engineering team develops practical AI, automation and integration capability for legal practice and business operations.

It works closely with Product, Data, Architecture, Security and operational teams to turn valuable ideas into secure, maintainable services, while retaining space for controlled experimentation and rapid learning.

The Role

Reporting to the AI & Innovation Lead, the Applied AI Engineer builds and improves AI-enabled products and workflows.

The role focuses on agents, model and tool selection, prompt and context engineering, retrieval, MCP, evaluation, token economics and the practical skills needed to make AI useful in legal and business work.

The role will work closely with the Product Manager, Automation & AI Engineer, lawyers, business teams, Data, Architecture, Security and vendors.

It will move use cases from rapid experiment and proof of concept into tested services, with clear evidence of quality, safety, adoption and value.

Duties And Responsibilities

• Applied AI delivery
• Work against the prioritised AI and innovation roadmap, with the Product Manager accountable for product priorities and outcomes and the AI & Innovation Lead accountable for innovation direction and team capability.
• Build AI-enabled applications, agents and workflow components using Python, Type Script, Azure and approved model services.
• Develop prompts, system instructions, tool definitions, retrieval approaches and context strategies suited to real legal and operational tasks.
• Design and evaluate retrieval pipelines, including document preparation, chunking, metadata, permissions-aware access, source attribution and retrieval quality.
• Build MCP integrations and other APIs that allow agents to use approved data and tools through controlled interfaces.
• Use AI-assisted software development to accelerate prototyping, while reviewing, testing and hardening generated code before release.
• Evaluation, quality and token economics
• Create representative evaluation sets from real use cases and define acceptance thresholds before wider release.
• Evaluate end-to-end outcomes and intermediate agent behaviour, including tool selection, task adherence, groundedness, safety and failure modes.
• Measure token consumption, latency and cost; compare models and architectures and recommend the simplest approach that meets the need.
• Use telemetry and user feedback to improve prompts, retrieval, skills, agents and model configuration after release.
• Track task-success rate, output quality, human correction or override, latency, unit cost, adoption and time or capacity created.
• Experimentation and productionisation
• Work with Product to turn opportunities into clear hypotheses, technical spikes, Po Cs and controlled pilots.
• Document experiment design, limitations, decisions and results so learning can be reused across the roadmap.
• Retain engineering ownership as successful prototypes move into secure, monitored and supportable production services, working with the Automation & AI Engineer on architecture, integration and production standards.
• Contribute reusable components, evaluation harnesses, prompt patterns and engineering guidance.
• Responsible delivery
• Protect confidential and personal data through appropriate access, retention and environment controls.
• Design human review, fallback and escalation into AI workflows where quality or consequence requires it.
• Explain model limitations and uncertainty clearly, avoiding the presentation of experimental capability as production-ready.
• Monitor production AI services for quality, cost, latency, tool failures and unexpected behaviour; investigate incidents and maintain appropriate runbooks, rollback arrangements and operational guidance.

Essential

Skills and Experience

• Strong software engineering fundamentals, including version control, testing, code review, documentation and secure development.
• Experience working with APIs, integrations, cloud services and event-driven or workflow-based solutions.
• Ability to move from an unclear problem to a working prototype, then improve it using evidence and user feedback.
• Confidence working with Product, Data, Architecture, Security and business specialists in a multidisciplinary environment.
• Clear written and verbal communication, including the ability to explain technical trade-offs to non-technical stakeholders.
• A pragmatic approach to experimentation, with good judgement about what is safe to test and what is ready to operate.
• Use of approved identity, secrets-management and service-connection patterns, avoiding embedded credentials and unmanaged service accounts.
• Commitment to accessible and usable user-facing applications and workflow interfaces.
• Hands‑on experience building LLM applications or agents, including prompts, tool use, retrieval and structured outputs.
• Strong Python capability and working knowledge of Type Script or Java Script.
• Practical experience evaluating generative AI systems using test data, traces and measurable quality criteria.
• Understanding of model selection, context windows, token consumption, latency and consumption-based cost.
• Curiosity and technical depth across agents, skills, MCP and emerging AI engineering patterns.

Desirable

• Experience in legal services, professional services or another regulated, knowledge-intensive environment.
• Experience with Azure, Git-based delivery, CI/CD, telemetry and production support.
• A strong portfolio of working software, prototypes or automation is more important than a specific degree.
• Experience with Microsoft Foundry, Azure Open AI, vector search, agent frameworks or Copilot Studio.
• Experience applying AI to document-heavy, high-stakes or professional workflows.
• Familiarity with model risk, content safety and human-in-the-loop controls.

Please note that this job profile is not an exhaustive list of duties but merely an outline of the key components of the role.

You may be required by your line manager to take on additional responsibilities when requested.

• About
• About Mishcon de Reya Group

The Mishcon de Reya Group is an independent, international professional services business with law at its heart, employing over 1400 people with over 650 lawyers.

It includes the law firm Mishcon de Reya LLP and a collection of leading consultancy businesses that complement the firm's legal services.

Mishcon de Reya LLP is based in London, Oxford, Cambridge, Singapore, Hong Kong and UAE.

The firm services an international community of clients and provides advice in situations where the constraints of geography often do not apply.

The work the firm undertakes is cross-border, multi-jurisdictional and complex, centred around three increasingly entwined and connected sectors: the Innovation Economy, Private Wealth and Capital, and Real Estate.

The firm is known as a disputes powerhouse with a formidable capacity firmwide for dispute resolution.

The Mishcon de Reya Group includes consultancy businesses MDR Discover , MDR Mayfair (in London, Singapore and Dubai), MDR ONE , and MDRi (in Hong Kong).

The Group also includes MDR Lab , which invests in the most promising early stage legaltech companies as well as the Mishcon Academy, its in-house place of learning and platform for thought leadership.

In 2024, the Group announced its first strategic acquisition in the alternative legal services market, flexible legal resourcing business Flex Legal.

It also acquired a majority stake in Somos, a global group actions management business.

We strive to create a fully diverse and inclusive workplace where all our people are empowered to fulfil their potential.

We are proud of our agile working culture and are always happy to talk flexible working.