Principal Applied Scientist
Upstart · United States · 2 wk ago
RemoteRemoteEducation$238k–$330k/yrFull-time
About the role
The Role:
- Define the technical vision for how offer decisioning systems should interconnect across partnerships, always-on systems, and marketplace optimization
- Build and guide conversion modeling approaches that optimize decisions across multiple stages of the customer journey rather than in isolated local steps
- Ensure models and decision policies at one stage account for downstream impacts, business constraints, and later-stage optimization opportunities
- Design interfaces between decision systems and optimization or constraint-specification components
- Drive cross-functional technical alignment across teams that currently own adjacent pieces of the problem
- Scope and lead large, ambiguous initiatives that require both deep modeling judgment and strong systems thinking
- Partner with scientists, engineers, and cross-functional stakeholders to translate analytical insights into durable production approaches
- Help ensure ongoing work across domains progresses toward a unified architecture and decisioning strategy
- Contribute directly to implementation and experimentation efforts, including prototyping models, reviewing code, and helping teams operationalize new approaches in production environments.
Qualifications
- Advanced degree in a quantitative field such as statistics, mathematics, economics, computer science, operations research, or a related discipline
- 8+ years of experience building and deploying machine learning models into production at scale
- Experience with optimization, operations research, or constrained decision-making problems
- Working knowledge of causal inference or causal machine learning
- Strong grounding in statistics and probability
- Experience leading large cross-functional technical initiatives with multiple stakeholders
- Experience working across multiple technical teams to align approaches, define interfaces, and move toward a shared vision
- Experience solving real-world machine learning or data science problems in a high-impact production environment