3 days ago
London, United KingdomSenior
Responsibilities
- Own AI features from ambiguous problem definition through reliable, observable production systems.
- Build retrieval and data foundations and treat retrieval quality as a core engineering problem.
- Design evaluation and monitoring approaches for quality, reliability, and safety.
- Build AI applications and agents using LLMs, orchestration, APIs, and data systems.
- Make and document decisions about models, operating cost, and latency.
- Prototype quickly, identify reusable patterns, and harden them into shared components.
- Improve engineering quality through code reviews, design feedback, and technical standards.
- Explain technical decisions to product managers and stakeholders in terms of business consequences.
Requirements
- Experience shipping AI or ML systems that real users depend on and understanding their production failure modes.
- Strong Python skills and practical SQL experience for assembling grounding data, building golden datasets, and investigating failures.
- Software engineering experience including testing, error handling, CI/CD, observability, and MLOps practices.
- Hands-on experience with LLM APIs, retrieval and vector stores, tool-using agents, and structured outputs on cloud infrastructure.
- Ability to evaluate model failures such as hallucination and prompt injection and assess their consequences.
- Ability to choose appropriately between models and simpler solutions and to stop when appropriate.
- Constructive code review and collaboration skills.
Benefits
- Competitive salary.
- Equity scheme.
- 25 days of holiday.
- Flexible working with 2 days per week in the London office.
- Private medical insurance via BUPA.
- Life assurance.
- Pension scheme with a 5% employer contribution.
- Enhanced family leave, including 26 weeks of full pay for maternity or adoption.
- 24/7 Employee Assistance Programme.
- EV leasing scheme.
- Cycle to work scheme.
- Nursery salary sacrifice scheme.
- One paid volunteering day per year.
Tech Stack
Categories
