Job description
Do you want to build production-grade AI agents with real commercial impact?
Have you taken LLM systems from prototype to secure, scalable production?
Are you ready to shape the ML foundations of an AI-first insurer?
A digitally-led UK specialist insuretech (200–300 employees) is undergoing a major AI transformation following recent refinancing and dedicated AI investment. A new Head of AI joined in December 2025 with a roadmap to scale the team to 10 and deliver end-to-end agentic automation across the claims lifecycle. Early AI products are already live, and this is a foundational hire in that journey.
This is an Machine Learning Engineer role focused on making AI capability production-ready, scalable and stable. You’ll lead the engineering of AI agents, influence architecture and define ML engineering standards across the platform.
Key Responsibilities
• Build and operate production-grade LLM-powered agents on GCP
• Design and scale robust RAG pipelines (ingestion, chunking, vector search)
• Integrate ML systems with policy, claims and CRM platforms via REST APIs
• Implement evaluation frameworks, monitoring, A/B testing and cost/latency dashboards
• Embed security, GDPR compliance, RBAC/IAM and auditability into ML systems
• Partner with Product and Claims to deliver measurable automation and cost savings
Key Details
• Salary: £80k-£90k base
• Working model: Hybrid – 1–2 days per week in South West London
• Tech stack: Python, GCP (Vertex AI, BigQuery, Cloud Run/Functions), LLMs, embeddings, vector databases, Docker, Kubernetes, CI/CD
• Visa sponsorship: Not available
Interested? Please apply below.
Extra information
- Status
- Open
- Education Level
- Secondary School
- Location
- London Area
- Type of Contract
- Full-time jobs
- Published at
- 07-03-2026
- Full UK/EU driving license preferred
- No
- Car Preferred
- No
- Must be eligible to work in the EU
- No
- Cover Letter Required
- No
- Languages
- English
Get similar vacancies sent to your mailbox
Fill in below which area you are searching in for a similar function and don't forget your e-mail address!