Research Engineer, Data & Evaluations for AI Legal Tech

Harvey AI; Greater London; On-site; GBP 85,000 - 120,000; Full time

Job summary

Harvey is transforming how legal and professional services operate with frontier AI and an enterprise-grade platform. We’re building data and evaluation systems powering post-training benchmarks and training signals for agentic workflows.

You’ll join a small, high-bar team of engineers and researchers owning large parts of the work, collaborating with researchers and legal experts to drive measurable improvements in model quality and product readiness.

Qualifications

• Experience building data pipelines, ML evaluation systems, or experimentation tooling.
• A rigorous approach to measurement, including statistical reasoning and experimental design.
• Ability to translate ambiguous goals or expert judgment into structured datasets, rubrics, and measurable criteria.
• Experience with synthetic data, LLM evaluations, automated raters, or preference data is a plus.

Responsibilities

• Build synthetic data pipelines that generate, filter, and curate training and evaluation samples for complex legal reasoning and agentic workflows.
• Design evaluations and verifiers that reflect real legal work and calibrate against expert judgment.
• Build and maintain capability benchmarks like the Legal Agent Benchmark (LAB) and product-focused evals for real customer workflows.
• Collaborate with legal domain experts to define quality metrics, rubrics, and grading mechanisms.
• Own training and evaluation data quality, including provenance, coverage, and safeguards.
• Close the loop from product usage and customer feedback to new evaluations and training data.
• Partner with researchers to investigate model failures and guide training and release decisions.

Skills

Python

Data pipelines

ML evaluation

Experimental design

Statistics

Job description

Harvey is transforming how legal and professional services operate with frontier AI and an enterprise-grade platform. We’re building data and evaluation systems powering post-training benchmarks and training signals for agentic workflows.

You’ll join a small, high-bar team of engineers and researchers owning large parts of the work, collaborating with researchers and legal experts to drive measurable improvements in model quality and product readiness.