Data Scientist; Insurance Risk and Pricing

Porch Group is a leading vertical software and insurance platform and is positioned to be the best partner to help home buyers move, maintain, and fully protect their homes. We offer differentiated products and services, with homeowners insurance at the center of this relationship. We differentiate and look to win in the massive and growing homeowners insurance opportunity by
1) providing the best services for home buyers,
2) led by advantaged underwriting in insurance,
3) to protect the whole home. As a leader in the home services software-as-a-service (“SaaS”) space, we’ve built deep relationships with approximately 30 thousand companies that are key to the home-buying transaction, such as home inspectors, mortgage companies, and title companies. In 2020, Porch Group rang the Nasdaq bell and began trading under the ticker symbol PRCH. We are looking to build a truly great company and are JUST GETTING STARTED.

The future is bright for the Porch Group, and we’d love for you to be a part of it as our Staff Data Scientist, Insurance Pricing. As one of the most senior individual contributors on our data science team, you will set technical direction for how Porch approaches homeowners insurance pricing and risk modeling, owning our most complex and ambiguous modeling problems end-to-end — from business strategy through production deployment.

While GLM-based pricing sits at the core of this role, your influence will extend across profitability and retention modeling, geospatial risk analysis, and the adoption of emerging techniques like generative AI. You’ll partner closely with actuarial, product, and engineering leaders to translate Porch’s unique property data into a durable competitive advantage in underwriting and pricing. This is an ideal role for a seasoned practitioner who wants to shape strategy, mentor other data scientists, and drive measurable business value at scale.

• Set the technical vision for insurance pricing and risk modeling, establishing best practices and modeling standards across the team
• Provide technical oversight and mentorship to other data scientists
• Architect and deploy GLM-based models (frequency, severity, and loss cost) for homeowners insurance pricing
• Build machine learning models (GBMs, neural networks, etc.) that drive underwriting accuracy, competitive positioning, and profitability
• Develop ensemble models predicting insured-level profitability, customer retention, and conversion, including customer lifetime value (LTV) models to prioritize marketing and underwriting strategies
• Lead use of non-traditional data sources — aerial imagery, satellite data, government records, building permits — to quantify localized risk and inform strategic decisions
• Partner with product, actuarial, engineering, and business leaders to scope high-priority initiatives and integrate data science solutions into operational workflows
• Work with the actuarial team to develop, file, implement, and monitor new predictive models that meet regulatory requirements
• Champion rigorous deployment practices in high-traffic environments, including A/B testing, performance monitoring, and continuous refinement
• Drive a culture of experimentation, evaluating emerging techniques including generative AI and LLMs to identify new opportunities for competitive advantage

• 10+ years of experience in data science, with significant depth in insurance pricing or risk modeling
• Track record of technical leadership — setting direction on complex projects, establishing standards, and mentoring or providing technical oversight to other data scientists
• Demonstrated expertise architecting, validating, and deploying GLM-based pricing models in production, ideally for homeowners or other property/casualty lines, as well as machine learning models for non-pricing use cases
• Proficiency in Python and SQL, with experience implementing GLMs and gradient boosting models (scikit-learn, stats models, xgboost, lightgbm)
• Experience with experimental design, including building, deploying, and A/B testing…