AI Engineer, Senior Associate in London

We’re looking for an AI Engineer who can take end‑to‑end ownership of features: from wrangling messy data, to building robust RAG pipelines, to shipping reliable APIs into production. You’ll sit between “wrapper dev” and “researcher”: comfortable with Python, data, and modern LLM tooling, and sceptical enough about AI outputs to keep our systems safe and accurate. Ideally, you have experience in agentic frameworks, some software development and strong desire to learn.

Key responsibilities include:

• Design, build, and maintain Retrieval‑Augmented Generation (RAG) pipelines over unstructured data (PDFs, HTML, emails, transcripts, APIs) using embeddings, vector databases (e.g. Pinecone, Weaviate, Qdrant), and Graph Databases (PuppyGraph, Neo4J, TigerGraph, ArangoDB, etc).
• Implement and tune chunking strategies to preserve context and improve retrieval quality, rather than naïve fixed‑length splits.
• Integrate LLMs (OpenAI, Anthropic, open‑source) via SDKs/HTTP, handling context windows, rate limits, retries, timeouts, and graceful degradation.
• Experience with AI coding tools: Cursor, Windsurf, etc. Someone who can look to remove the ‘vibe’ from ‘vibe coding’, and call out AI when it produces poor code, based on their human coding experience.
• Leverage AI-assisted development (e.g., Cursor/Windsurf/GitHub Copilot/LLM agents) to create production‑quality software faster – generating scaffolding, tests, and documentation – while applying rigorous human review for correctness, security, and maintainability.
• Designing multi‑agent / “agentic” workflows where specialized AI agents coordinate (triage, research, drafting, review) to solve complex tasks.
• Implementing prompt‑injection defences, output filtering, role/permissioning, and safe tool‑use patterns.
• Knowledge of data privacy and governance concerns around AI (GDPR, SOC2) and experience with dataset auditing / fairness evaluation is a plus.
• Experience fine‑tuning or adapting open‑source models (e.g. Llama, Mistral) and managing training/inference pipelines.
• Build ETL jobs and ingestion scripts to clean, normalize, and enrich text data (Python, pandas, BeautifulSoup or equivalent).
• Work with both SQL and vector stores; design schemas and indices that support low‑latency semantic and hybrid search.
• Design, iterate, and version prompts (system, user, tool) using techniques like few‑shot examples and chain‑of‑thought to improve reliability on complex tasks.
• Own evaluation (evals) for your features: create test sets, define success metrics (accuracy, faithfulness, latency), and run regression tests before and after changes (e.g. Ragas or custom eval harnesses).
• Monitor production behaviour, debug hallucinations, and systematically reduce failure modes through better retrieval, prompting, and guardrails (not just “tweak the temperature”).
• Build and maintain typed, well‑tested backend services (e.g. Python with FastAPI/Flask; Node/TypeScript as a plus) that expose AI capabilities to front‑end and internal consumers.
• Implement observability for your services (logging, metrics, tracing, token usage) and contribute to dashboards for reliability and costs.
• Work closely with product, design, and ops to scope AI features, translate fuzzy requirements into concrete technical plans, and iterate based on user feedback.
• Take end‑to‑end ownership of projects: from prototype through to production, maintenance, and continuous improvement.
• Be responsible for ensuring all information security processes, policies and procedures are adhered to and any issues or concerns are raised with the Cyber Security team.
• Ensure full compliance with all local data protection regulations and privacy controls, and any related issues are raised via the appropriate channels.

Key Requirements:

• Keeps up to date with the latest developments in AI and loves to experiment with new models.
• Bachelor’s or Master’s degree (preferred) in Computer Science, Artificial Intelligence, Data Science, or a related field – including formal AI/ML study (e.g., machine learning, deep learning, NLP, statistics) or equivalent professional training/portfolio.
• Nice to have: Experience with Microsoft Power Platform (Power Automate, Power Apps, Dataverse) and/or Azure Logic Apps for workflow automation and integrating AI services into business processes.
• Previous experience in investments/ finance sector or autonomous systems is a plus.
• Experience in: Natural Language Processing (NLP), Computer Vision, Reinforcement Learning, Generative AI (e.g., GPT, Stable Diffusion).
• Approximately 3+ years professional software engineering experience, including 1–2 years building production AI/ML or LLM‑based systems. (Strong non‑traditional backgrounds with equivalent portfolio are welcome.)
• Expert proficiency in Python for backend and data work (typing, async, packaging, testing, performance profiling); familiarity with TypeScript/Node/React is a strong plus.
• Hands‑on experience with at least one orchestration framework (e.g. LangChain, LlamaIndex, etc.).
• Practical experience implementing RAG: embeddings, vector databases (Pinecone, Weaviate, Milvus, Qdrant, etc.), and semantic search.
• Comfortable with SQL and working with pandas/DataFrame‑style data manipulation.
• Experience building and deploying APIs/microservices (REST, JSON, auth, rate limiting, pagination, error handling).
• Nice to have: Strong understanding of AI infrastructure for scalable model serving, distributed training, and GPU orchestration.
• Solid software engineering fundamentals: testing, code reviews, version control (Git), CI/CD, and basic cloud services (AWS/GCP/Azure).
• Nice to have: Experience with IAC tools like Terraform and Crossplane.
• Rapidly prototype AI solutions, test emerging tools, and recommend best practices for adoption.
• Knowledge of MCP (Model Context Protocol) is a plus.
• Ability to explain complex AI concepts simply to stakeholders.
• Working understanding of how LLMs behave in practice: context windows, tokens, temperature/top‑p, hallucinations, and prompt injection risks.
• Familiarity with core concepts: embeddings, vector similarity, Transformers at a conceptual level (you don’t need to derive attention, but you should know what it does).
• Demonstrated experience shipping production AI feature (e.g. RAG chatbot, summarization/search assistant, agentic workflow, etc.). Portfolio or GitHub strongly preferred.
• Strong problem-solving skills with ability to translate business needs into AI solutions.
• Evolve with AI – you realise we are universal learners, always and forever improving and pushing the next frontier.
• ‘Fail fast’ – looking to have someone who experiments quickly and learns even quicker!
• “AI scepticism”: you use AI tools for speed but verify outputs and design systems assuming models will sometimes be wrong.
• Product thinking: you care about UX quality, not just whether the API returns 200.
• Ability to work with ambiguity, learn new tools quickly, and keep up with a fast‑moving AI ecosystem.

Partners Capital is committed to being a great place to work. We are focused both on wellbeing and professional growth. You can expect professional development and career progression opportunities, competitive compensation, exceptional benefits and a flexible “results-focused” working model. Our benefits package includes private medical and life insurance, income protection and pension contributions. In addition, we partner with organisations to provide wellness benefits. Partners Capital supports global philanthropy via a charity program and provides a volunteer day for all employees. We also champion a variety of social events.