We develop AI-assisted legal research tools for law firms and in-house legal teams. Our clients handle high-stakes matters where an imprecise output — a hallucinated citation, an incomplete clause extraction — carries real professional risk. We need a prompt engineer with the discipline to work in this environment: systematic testing, documented version control, regression suites, and close collaboration with practising lawyers to validate accuracy. Generalist prompt engineers will not be a fit here.
Responsibilities
Design and test prompts for legal document summarisation and clause extraction
Build a regression test suite to catch prompt quality regressions
Document all prompts with version history and performance benchmarks
Work with lawyers to validate output quality and refine prompt behaviour
Monitor production prompts and respond to quality incidents
Requirements
Demonstrated experience prompt engineering for high-stakes or regulated domains
Systematic approach to prompt testing and output evaluation
Familiarity with structured output techniques (JSON mode, tool use, function calling)
Able to build a prompt regression test suite
Interest in or familiarity with legal language is a strong plus