We built 34 open-source Python libraries for UK personal lines pricing — causal elasticity estimation, SHAP-based factor extraction, fairness auditing, conformal prediction, Databricks deployment pipelines. They get 17,200+ monthly PyPI downloads and are used by pricing teams across the UK. We are practitioners; the libraries are the work record.
Pricing model review
We review your frequency and severity model stack against FCA Consumer Duty requirements and PRA SS1/23. That means temporal cross-validation correctness, calibration by segment, IBNR handling, feature engineering decisions, and the governance documentation a model risk function will accept. You receive a written assessment and a ranked issue list within two weeks.
Databricks deployment
We deploy our open-source toolkit on your Databricks workspace, configured against your data schema, claims conventions, and MLflow setup — including champion/challenger deployment with the audit trail ICOBS 6B.2 requires. Typically three to five days.
Team training
We deliver the Modern Insurance Pricing course to your team using your data and your models. Not generic Python training: CatBoost for frequency and severity, SHAP multiplicative factor extraction, temporally correct cross-validation, governance documentation. Typically two days on-site.
Typical engagement: £5,000–£25,000. Fixed price, scoped upfront.
Email pricing.frontier@gmail.com with “Consulting enquiry” in the subject.