London AI Engineer

Job description

Job Description Want to shape how one of the world's leading publishers builds with AI? Join Penguin Random House as our first AI Engineer, and help build the AI-powered products that are changing how we understand, publish and sell books. About the team Our Data Science and Analytics team builds data-driven tools that help us make smarter, ethical and commercially responsible decisions across our books. As our work has grown in scale and ambition, AI has transformed what we build and how quickly we can build it. We're now creating AI-powered applications that make teams faster and crack business problems that were out of reach only a few years ago. We're also part of Penguin Random House's global community of AI and ML engineers, working with colleagues in the US and around the world to share tools, compare notes and shape how data science and AI are done across the company. About the role As our AI Engineer, you'll be at the heart of this work, bringing together strong software engineering, a solid grounding in machine learning and hands-on generative AI experience. You'll design and build language-model workflows for summarisation, extraction, classification and search, and add natural-language interfaces and conversational search to the products our colleagues use every day. You'll also own the AI models we rely on. That means testing, deploying and caring for them over time, so that we choose models on evidence and keep costs in check. Beyond the products themselves, you'll lead the team's use of AI coding assistants, setting the standards, shared skills and workflows that let us move fast without cutting corners. You'll work hand in hand with our ML Engineers and ML Infrastructure Engineers: they build the models and the platform, and you build the AI products on top. You'll also be the go-to person for prompt and agent development, teaming up with colleagues across Sales, Marketing and Editorial to turn their ideas into solutions that scale, using AI where it fits best and saying so when it doesn't. If you'd rather build AI that's useful, reliable and responsible than AI that simply impresses in a demo, and you love helping the people around you level up, we'd love to hear from you. Key responsibilities: Design, build and improve language-model-based workflows for tasks such as classification, summarisation, extraction, generation and search, and develop the Python services, APIs and internal tools that integrate these into day-to-day business workflows Own model evaluation across projects: design and run the testing of new AI models and providers, maintain evaluation frameworks and datasets, and make evidence-based recommendations that balance quality, speed, reliability and cost Own the AI model lifecycle: manage the deployment, configuration, ongoing maintenance, upgrade and retirement of the AI models in use by the team, working with ML Infrastructure Engineers on gateway, serving and observability Monitor AI usage and cost across our applications, identify edge cases and failure modes, and improve system behaviour and reliability Work with teams across Sales, Marketing, Editorial and the broader business to translate their needs into clear, scalable solutions, directing our effort and applying AI where it best fits – and advising when it does not Lead AI code assistance for the team: develop and maintain common standards, skill files, agent configurations and workflows; establish best practice for the use of AI tools; and ensure the team is upskilled with the latest developments Liaise with our US and global colleagues to share knowledge and tooling, and ensure we keep pace with them Support other teams with prompt development and agent development where this intersects with our remit, turning today’s ad-hoc help into a repeatable service Establish and follow best practice for responsible use of AI, including data privacy, provenance and appropriate human oversight Stay current with the rapidly evolving AI landscape and help the team evaluate which tools, models and approaches are worth adopting What you’ll bring Essential criteria: Expert Python user with strong software engineering fundamentals: testing, typing, packaging and code review Experience designing and building maintainable APIs and full-stack applications with Python and TypeScript (FastAPI, React or similar), including front-end prototyping Demonstrated experience building LLM-based or applied AI applications in production, including prompt engineering, structured outputs, tool use and function calling, RAG pipelines and agentic workflows Experience evaluating LLMs, including building evaluation datasets and using LLM-as-judge techniques appropriately Deep, day-to-day experience with agentic coding assistants (Claude Code, Cursor, GitHub Copilot, Codex or similar) Ability to understand what stakeholders need and turn it into effective solutions, explaining technical subjects in plain language and saying clearly when AI isn't the right answer Proven ability to manage priorities well in a hybrid working environment Preferred criteria: Experience with LLM and agent frameworks (PydanticAI, DSPy, GEPA, LangChain/LangGraph, provider agent SDKs or similar), and an informed view of when to use them and when not to Experience with evaluation and observability tooling such as MLflow, LangSmith or Arize Experience configuring agentic coding assistants for a team, with shared instructions, skills and workflows Solid grounding in machine learning and NLP: able to reason about model behaviour, understand embeddings and retrieval, and know when a classical ML approach beats a language model Working knowledge of the major model providers and their APIs, and of the trade-offs between hosted and self-hosted models in quality, latency, cost and data handling Strong SQL and comfort working with unstructured text data at scale Familiarity with CI/CD and containerised deployment: Docker, Kubernetes and cloud platforms (AWS preferred) Awareness of security and privacy considerations specific to AI applications: prompt injection, data leakage and handling of sensitive content A natural teacher and advocate, with experience mentoring colleagues and raising a team's skills with AI tools Application instructions Please apply with your CV and cover letter (both in pdf format) outlining why you are the right candidate for the role by 11:59pm on Thursday 22nd October. The cover letter should be uploaded as an additional document. Please ensure you include a cover letter (in pdf format), as it is a crucial part of our assessment process. The cover letter offers an opportunity to show how your experience and interests align with the role requirements. Typically, we expect the cover letter to be no more than one or two pages in length. Salary Circa £55,000 dependent on how your skills and experience align to the role, plus bonus and benefits. AI Here at Penguin, we believe in the power of authenticity and human creativity. When you apply for a position, we want to encourage you to showcase your unique voice. Throughout our recruitment process, please share your own thoughts, experiences, and skills. This helps us get a true sense of who you are and what you might bring to our team. We celebrate creativity and diverse perspectives, so please be yourself! While we recognise AI tools can be helpful, we recommend using them thoughtfully to ensure your responses reflect you. Disability Confident As a Disability Confident Committed organisation, we offer interviews to candidates with a disability who meet the essential criteria for the role, and opt-in on their application form. The essential criteria for this role are listed as part of the ‘What you’ll bring’ section. There may be times when the volume of applications means we cannot take all eligible candidates to interview. We encourage you to tell us about any reasonable adjustments you may need by emailing PRHCareersUK@penguinrandomhouse.co.uk. Remember, y