Overview

Storytelling is, and will always be, the heart of what we do at Skydance.

Responsibilities

• Lead end-to-end development of production ML systems: data sourcing, feature engineering, model training, evaluation, deployment, and monitoring.
• Own key ML products such as probabilistic identity resolution (matching unauthenticated device IDs and 1P cookies to households/persons with calibrated confidence), single-title affinity (e.g., STAT two-tower retrieval), and audience/propensity models.
• Design scalable feature pipelines on Databricks (PySpark, Delta, Workflows/DLT, Unity Catalog) and the WBD feature store, with documented feature contracts, backfill paths, and freshness SLAs.
• Architect batch and near-real-time inference pipelines integrated with Snowflake and activation systems (Mosaic, FreeWheel, GAM).
• Develop and optimize models across the ML spectrum: gradient boosting (XGBoost/LightGBM), embedding/two-tower retrieval, neural ranking, probability calibration (e.g., isotonic regression), and probabilistic/graph-based matching.
• Design rigorous offline and online experiments; define evaluation frameworks (precision/recall, AUC-ROC, NDCG, decile lift, calibration curves) appropriate to each use case.
• Apply causal-inference techniques (propensity scoring, uplift/incrementality modeling) to measure true lift of audience targeting on engagement and retention KPIs.
• Contribute to lookalike modeling (LAL 2.0+) using 1,000+ first- and third-party features, including privacy-safe builds inside Data Clean Rooms (Snowflake DCR).
• Champion MLOps best practices: model versioning, champion/challenger promotion, automated retraining triggers, drift detection, and production monitoring with MLflow on Databricks.
• Build and maintain robust, reproducible, auditable ML pipelines on Databricks (and AWS SageMaker where appropriate, e.g., the identity-resolution track); enforce leakage prevention and training/serving consistency.
• Contribute to the team’s feature-store strategy — feature contracts, backfills, and freshness SLAs — and implement data-quality checks, model-health dashboards, and alerting thresholds.
• Embed FinOps cost discipline (compute caps, auto-termination, job tagging) into pipeline design.
• Mentor MLE 2s through code reviews, design discussions, and pairing; contribute to team technical standards.
• Partner with Product, Marketing, and Ad Sales to translate business requirements into ML problem formulations, and with Data Engineering on data contracts and pipeline SLAs.
• Communicate model performance, trade-offs, and business impact clearly to technical and non-technical stakeholders.
• Identity Intelligence — foundational, privacy-safe identity across all WBD brands: probabilistic ID resolution that resolves unauthenticated signals to households/persons with calibrated confidence (entity resolution with gradient boosting and embeddings, representation learning, isotonic calibration, candidate blocking, champion/challenger pipelines), expanding addressable audiences beyond deterministic matching.
• Audience Intelligence — advertising and marketing use cases: lookalike and predictive audiences (LAL across 1,000+ features), ML-driven smart audiences, layered retrieval + propensity, and incrementality/closed-loop optimization, with privacy-safe activation including data clean rooms.
• ML-based Forecasting — audience growth, demand, and advertising yield/pricing forecasting that powers ad sales and marketing decisions.
• Content Preferences & Affinity — genre-preference, content-preference, and single-title affinity modeling (two-tower retrieval with semantic content embeddings) that ranks audiences for upcoming titles and powers cross-channel promotion.

Qualifications

• 5–8 years of industry experience in ML engineering or applied data science (3+ years with a Ph.D.).
• Deep Python expertise and production-quality software engineering practices; production experience building and deploying ML at scale (millions+ of users/records).
• Strong proficiency in Databricks (PySpark, Delta Lake, Workflows/DLT, MLflow, Unity Catalog) and solid SQL/Snowflake experience for feature sourcing and model-output delivery.
• Experience with AWS ML services (SageMaker, S3, Lambda).
• Strong understanding of ML model evaluation, A/B testing, and statistical inference; knowledge in one or more of recommendations & ranking, identity resolution, embeddings/retrieval, causal/interpretable ML, forecasting, bandits, or optimization.
• Demonstrated ability to lead technical decisions and mentor engineers.
• Bachelor’s or Master’s degree in Computer Science, Statistics, Engineering, or a related quantitative field (or equivalent experience).
• Excellent written and verbal communication, with the ability to advocate technical solutions to engineers, scientists, and product stakeholders.
• Recommendation systems, personalization, identity resolution, or audience modeling in a media / streaming / ad-tech context.
• Familiarity with Data Clean Room environments (Snowflake DCR, AWS Clean Rooms) and consent-aware activation.
• Experience with two-tower / retrieval architectures, probabilistic identity resolution (graph-based matching, entity resolution), and probability calibration.
• Hands-on experience with agentic AI frameworks (LangChain, LangGraph, AutoGen, MCP), Databricks Genie Space configuration, and Snowflake Cortex.
• Experience with feature stores (Databricks Feature Store, Tecton, Feast) and contributions to open source or ML publications.

Benefits

• A Great Place to work
• Equal opportunity employer
• Fast track growth opportunities