Senior Machine Learning Engineer

About the Role

This is a senior, high-ownership US-based role that sits between our Senior MLE and Staff MLE levels. You will own the design and delivery of production ML systems end to end and take on cross-cutting technical leadership: setting patterns, driving key architectural decisions on flagship workstreams, and raising the bar for the broader ML organization — including close partnership with our Hyderabad ML team. As a US-based senior engineer, you will also serve as a technical anchor and time-zone bridge across the global team: framing ambiguous problems, unblocking others, and translating business priorities from US-based Product, Marketing, and Ad Sales stakeholders into an executable ML roadmap. This role is ideal for engineers with roughly 5–8 years of experience (3+ with a PhD) who operate with strong autonomy, lead by influence, and can move fluidly from hands-on modeling and pipeline engineering to architecture and mentorship. You will do meaningful individual technical work while beginning to exercise Staff-level scope across initiatives.

Responsibilities

• ML System Design & Technical Leadership
  • Lead end-to-end development of production ML systems: data sourcing, feature engineering, model training, evaluation, deployment, and monitoring.
  • Own one or more flagship ML products — e.g., probabilistic identity resolution (matching unauthenticated device IDs and 1P cookies to households/persons with calibrated confidence), single-title affinity (two-tower retrieval), lookalike modeling, or forecasting — and drive their technical direction.
  • Make and document key architectural decisions across a workstream (feature-store design, training/serving patterns, evaluation frameworks); provide deep trade-off analysis on scalability, latency, reliability, and cost.
  • Design scalable feature and inference pipelines on Databricks (PySpark, Delta, Workflows/DLT, Unity Catalog) integrated with Snowflake and activation systems (Mosaic, FreeWheel, GAM), with documented feature contracts, backfill paths, and freshness SLAs.
  • Establish and evangelize patterns that other engineers adopt; anticipate risks and failure modes before they surface.
• Modeling & Experimentation
  • 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).
• MLOps & Infrastructure
  • 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.
  • Shape 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.
• Agentic AI & Modern Development
  • Actively use and advocate for AI-assisted development: Cursor, GitHub Copilot, and Amazon Q for code generation, review, and documentation.
  • Leverage Databricks Genie as a governed natural-language analytics layer — configuring Genie Spaces over ML feature tables and audience datasets to enable self-service exploration for cross-functional stakeholders.
  • Use Snowflake Cortex (Copilot, Cortex Analyst, Cortex Search) to accelerate SQL authoring, data discovery, and RAG-based internal tooling over Snowflake-resident identity and audience data.
  • Design and prototype agentic ML workflows (MCP-compatible tooling, LangChain/LangGraph) to automate repetitive tasks such as data validation, feature selection, and hyperparameter search; evaluate LLM-based approaches for metadata enrichment and content understanding.
• Mentorship & Cross-functional Collaboration
  • Mentor Senior and MLE 2 engineers — including members of the Hyderabad team — through code reviews, design discussions, and pairing; contribute to and help set team technical standards.
  • Serve as a US-based point of contact and time-zone bridge for the global ML team; help align priorities and unblock the India team across time zones.
  • Partner with US-based Product, Marketing, and Ad Sales stakeholders 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.

Flagship Projects

• 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.

Requirements

• 5–8 years of industry experience in ML engineering or applied data science (3+ years with a Ph.D.), including a track record of leading projects to production.
• Deep Python expertise and strong 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/causal inference; depth in one or more of recommendations & ranking, identity resolution, embeddings/retrieval, forecasting, or optimization.
• Demonstrated technical leadership: driving architectural decisions, setting patterns/standards, and mentoring other engineers — including leading by influence across teams and time zones.
• 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.

Preferred Qualifications

• Recommendation systems, personalization, identity resolution, or audience modeling in a media / streaming / ad-tech context.
• Experience with two-tower / retrieval architectures, probabilistic identity resolution (graph-based matching, entity resolution, confidence calibration), and Data Clean Room ML (Snowflake DCR, AWS Clean Rooms).
• Experience architecting or standardizing components of an ML platform used by multiple engineers or teams.
• 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.
• Experience partnering with or mentoring globally distributed teams.

Skills

Primary platform: Databricks (Lakehouse, PySpark, Delta, Workflows/DLT, MLflow, Feature Store, Unity Catalog, Asset Bundles, Genie).

Cloud: AWS (SageMaker, S3, Lambda).

Warehouse: Snowflake (incl. DCR, Snowpark, Cortex).

Activation: Mosaic, FreeWheel, Google Ad Manager.

Agentic AI: Cursor, GitHub Copilot, Amazon Q, Databricks Genie, Snowflake Cortex, MCP.

Languages: Python (primary), SQL, Scala (as needed).