Remote Founding AI Engineer: Causal Inference

Referment; Greater London; Remote; GBP 90,000 - 150,000; Full time

Benefits offered by this job

Founding-team equity

Job summary

Referment is building an applied AI research lab and seeks a senior researcher to own causal inference and experiments. You will map drivers of client numbers, estimate effects from natural experiments, and design client-focused experiments to translate forecasts into actionable ranges.

Join a small founding team, work with cutting-edge LLM tools, and shape how methods are delivered on the platform while maintaining rigorous evaluation standards and clear cause-and-effect explanations for

Qualifications

• 5+ years building AI/ML systems in production.
• Causal inference applied to real interventions with known limits.
• Strong statistics, uncertainty calibration and experiment design.
• Excellent Python (async, typed, tested) and SQL with cloud deployments.
• Experience with LLM tooling beyond demos (LangChain/LangGraph/LlamaIndex).

Responsibilities

• Build the causal layer of the platform and map drivers of client numbers.
• Estimate effects from natural experiments and validate estimates before use.
• Design experiments with clients and translate forecasts into ranges.
• Own the attribution method crediting each decision’s result.
• Design benchmarks, rubrics and blind holdouts for trustworthy evaluation.

Skills

Causal inference

Python

SQL

Experiment design

Statistics

Cloud (AWS/Azure)

LangChain

LangGraph

LlamaIndex

Education

Top-tier university degree

Tools

LangChain

LangGraph

LlamaIndex

Claude Code

Job description

Referment is building an applied AI research lab and seeks a senior researcher to own causal inference and experiments. You will map drivers of client numbers, estimate effects from natural experiments, and design client-focused experiments to translate forecasts into actionable ranges.

Join a small founding team, work with cutting-edge LLM tools, and shape how methods are delivered on the platform while maintaining rigorous evaluation standards and clear cause-and-effect explanations for