Actuarial Developer

Shepherd•San Francisco, CA

26d

About The Position

You will help design, build, and maintain the internal pricing service and data models that allows actuaries to deploy new python-based rating engines and allows insurance product heads to directly configure our appetite as Shepherd grows into new sectors, coverages, and lines of business. You will ensure Shepherd’s platform serves the actuarial and insurance product use case. In this role, you are not handing requirements over a wall to engineers. You are in the codebase, in the architecture discussions, in the PR reviews. The split is roughly: 30–40% of your time writing code yourself, and 60–70% designing systems, reviewing implementations, and guiding adjacent engineering decisions with actuarial context. This is an individual contributor role for someone with real actuarial depth and experience deploying versioned, well-tested specialty and commercial lines rating engines to production. If you've implemented hx Renew or a similar pricing platform at a carrier or MGU, this is the role for you. If you’ve spent your career wishing actuarial pricing could move at the speed of software this is the role for you.

Requirements

ACAS or near-ACAS exam progress. You have 4+ years of P&C commercial lines actuarial experience. You understand ISO rate plans, large account pricing concepts, experience rating, collateral, because you've built or maintained pricing models that have used them. We care about the judgment, not the letters, but exam progress is a good indicator of the former.
Production Python experience. You've written production python for 2+ years that other people depend on: pricing tools, data pipelines, internal APIs, analytical applications. Our stack includes EC2, Lambda, FastAPI, pytest, uv, dbt, dagster, postgres.
Experience designing or working within rating engine architectures. You know how commercial and specialty lines (e.g., CGL, Business Auto, Workers' Comp, Excess, Commercial Property) rate plans are structured and you have implemented them in python.
Previous titles might include: Pricing Developer, Actuarial Data Engineer, Pricing Actuary.

Nice To Haves

Experience with hx Renew or a similar Python-based pricing platform. You have either configured models on it or led its implementation at a carrier or MGU.
Data pipeline experience. You are familiar with third-party data sources used in commercial underwriting. You've built or maintained ETL/ELT workflows, wrangled messy external data, or supported ML/GLM model deployment. You don't need to be a data engineer, but you should be comfortable with orchestration concepts, columnar formats, and data quality tooling.
Hands-on predictive modeling for insurance pricing. GLMs, gradient-boosted trees, loss cost models, or similar approaches that produce risk scores feeding into traditional rating structures.

Responsibilities

Rating engine service (50-70%). You'll design and develop the FastAPI service, in collaboration with engineers, that allows the python-native actuarial team to implement rating logic for multiple lines of business that make account-level pricing coherent and extensible to new sectors & new coverages.
Product configuration service (20-30%). You will design and develop a service that enables Insurance Product heads to update product configurations (coverage options, pricing parameters). You'll design the data model, versioning scheme, and validation logic that makes this safe and self-service.
Data integration architecture (0-20%). You'll evaluate third-party data sources for pricing segmentation benefits and collaborate with data engineers on how best to integrate these data sources for the needs of predictive modeling, underwriting, and business development. You'll need to be comfortable with data quality validation and the messy reality of third-party commercial data.
Predictive models, occasionally (0-20%). Loss cost models trained on our claims corpus, increased limits factor analysis, and behavior-based pricing models that turn insureds’ operational data into quantitative risk signals.