Ofertas › Pacific Life
Pacific Life · London, United Kingdom · £35,000 - 43,000 per year
Salary: £35,000 - 43,000 per year Requirements: 6+ years in ML/AI, with deep hands-on expertise in NLP, LLMs/SLMs, and agentic systems. Hands-on experience building with OpenAI and Anthropic models/APIs; strong knowledge of LLMs and agent architectures. Practical experience with LiteLLM, vLLM, and Langfuse or equivalent LLM observability tooling; experience with Claude Code/Codex and associated coding-agent workflows is a plus. Proven ability to design, build, and own evaluation harnesses for LLM/agent outputs. Proven ability to design and optimise prompts and context, including techniques like few-shot, chain-of-thought, and prompt tuning, grounded in measurable evaluation rather than guesswork; prior experience with DSPy, GEPA, or similar is a plus. Strong Python skills, with FastAPI for service development, Postgres or similar for data persistence, and comfort with event-driven, async architectures. Working knowledge of ML frameworks such as PyTorch or TensorFlow, containerisation with Docker and Kubernetes, and cloud platforms, with AWS preferred. Entrepreneurial, ownership-driven, and biased to action, with a Build Fast, Fail Fast approach, thriving in agile, iterative environments. Responsibilities: Design, experiment, and deploy production LLM and agentic systems. Quickly test hypotheses, optimise, and iterate based on real-world feedback and evaluation results. Build and maintain the evaluation harness that tells us whether a model or prompt change made things better or worse. Architect multi-step, tool-using pipelines with explicit state, guardrails, and human-in-the-loop review where required. Push the envelope with cutting-edge LLM and agentic techniques to solve hard problems in language understanding, extraction, and reasoning. Collaborate with Product, Design, and Underwriting SMEs to align ML capabilities with product goals and real domain constraints. Take systems from prototype to production, optimising for latency, cost per run, and reliability at scale. Own features from idea through to production impact and focus on whether the system actually solves the underwriters problem. Technologies: AI AWS Architect Claude Code Cloud Docker FastAPI Support Kubernetes LLM Langfuse PyTorch Python TensorFlow vLLM Backend Machine Learning More: We are UnderwriteMe, an insurtech software business and subsidiary of Pacific Life Re, a global life and pensions reinsurance firm. Our vision is to help everyone purchase protection insurance by using data a