Staff/Senior Machine Learning Scientist - Forecasting (Open to Remote)

Penguin Random House US•New York, NY

18h•Remote

About The Position

Penguin Random House is the largest trade publishing company in the world. The Data Science team is seeking an experienced Machine Learning Scientist to drive business-critical forecasting products. We have a mature machine learning practice with strong infrastructure, supported by data warehouse and DevOps partners. We are transitioning to AI-accelerated development and use modern agentic coding tools like Claude Code to speed up how we build and maintain ML systems, with rigorous quality gates including tests, reproducible workflows, and measurable improvements in model performance and reliability. Experience with agentic workflows is a plus, but we prioritize strong fundamentals and the ability to learn new workflows effectively. This role may be filled at the Senior or Staff level depending on experience and interview performance.

Requirements

Senior level: 5+ years in applied ML/data science, including owning models in production (deployment, monitoring, incident response, retraining)
Strong forecasting expertise (time-series methods, feature engineering, rigorous backtesting) OR deep expertise in Bayesian statistical methods and probabilistic programming
Strong statistics fundamentals; comfort with probabilistic forecasting and explaining uncertainty in practical terms
Strong Python (or R) and SQL; writes production-quality, testable code
Strong communication and cross-functional collaboration with non-technical stakeholders
Experience using AI-assisted development workflows responsibly (verification loops, reproducibility, automated checks)
Staff level: 8+ years in applied ML/data science, or PhD with 3+ years of applied experience
Experience building ML systems end-to-end (not just models): backtesting frameworks, scheduled retraining, monitoring/alerting, and automated reporting into planning or decision workflows
Demonstrated ability to inherit complex systems built by others and make sound architectural decisions with high autonomy
Technical leadership: raises the bar on evaluation, reproducibility, and production practices; mentors less-senior team members

Nice To Haves

Experience with agentic workflows is a plus

Responsibilities

Own end-to-end ML systems: scoping, feature engineering, model development, backtesting/validation, deployment (with platform partners), monitoring/alerting, retraining cadence, and ongoing reliability improvements.
Create and maintain production-safe evaluation infrastructure: automated backtests, error decomposition, uncertainty quantification, data validation, regression gates, and auditable model/version lineage.
Build AI-assisted/agentic development workflows (e.g., Claude Code) to automate repetitive tasks with human review and measurable quality gates.
Define success metrics tied to business outcomes; communicate assumptions, limitations, and risk so model outputs are used correctly by stakeholders.
Write production-quality, testable code and support reproducible workflows.
Partner across functions to translate business needs into a prioritized technical roadmap and measurable impact.
Build and improve forecasts across time horizons and business segments (demand, inventory, supply chain, resource allocation), selecting approaches that balance accuracy, stability, interpretability, and operational cost.
Productize forecast outputs for stakeholders: clear definitions and assumptions, versioned releases, and reporting that explains what changed, why it changed, and how uncertainty should shape decisions.
Feature engineering, uncertainty quantification and calibration, hierarchical/segmented forecasting where appropriate.
Partner with operations, supply chain, inventory, finance, and marketing leaders.