Staff Data Scientist, Upstart

• Master’s degree in a quantitative field (e.g., Statistics, Mathematics, Economics, Finance, Operations Research, Physics, or related field)
• 6+ years of experience in analytics, data science, or a related quantitative field
• Proficiency in Python (Pandas/Numpy) and SQL
• Strong analytical and first-principles thinking, with the ability to tackle complex and ambiguous problems
• Strong communication, presentation, and collaboration skills, with the ability to clearly articulate complex analytical concepts to diverse audiences

Desirable

• As a Staff Data Scientist at Upstart, you will own the monitoring and performance analytics of our Personal Loans underwriting models and credit portfolio. This role provides regular visibility to senior leadership and plays a key role in shaping data-driven decisions at the highest levels of the organization
• In addition to leading recurring reporting and performance tracking, this role requires strong analytical curiosity and investigative rigor.
• You will proactively monitor portfolio and model performance, identify emerging trends, diagnose unexpected outcomes, and recommend actionable treatments.
• This role is well-suited for someone who enjoys going beyond standard reporting to conduct deep analytical investigations and drive meaningful business impact in a production machine learning environment
• Lead end-to-end monitoring of underwriting model performance and outcomes, ensuring models are behaving as expected across risk segments and macroeconomic conditions; translate model metrics into clear business implications
• Detect shifts in approval rates, loss performance, borrower behavior, or model predictions, and use analytical methods to investigate drivers of change, distinguish signal from noise, and quantify business impact
• Apply statistical techniques to assess A/B tests, model launches, and natural experiments, and provide clear recommendations grounded in causal reasoning and quantitative evidence
• Develop and maintain automated reporting, monitoring frameworks, and workflows that ensure timely, accurate visibility into model performance at scale