Machine Learning Engineer 5 - Ads Inventory Management & Forecasting

Netflix Netflix · Big Tech · Los Gatos, CA +3 · Data & Insights

Machine Learning Engineer at Netflix focused on Ads Inventory Management & Forecasting. The role involves building end-to-end ML model deployment and inference infrastructure for low-latency real-time ad systems, handling large data volumes with Spark, and productionizing predictive models for campaign effectiveness forecasting. It also includes building scalable simulation solutions for inventory scenarios and collaborating with cross-functional teams.

What you'd actually do

  1. Experience in building end-to-end ML model deployment and inference infra for low-latency real-time ad systems.
  2. Experience in handling data at extremely large volumes with big data tools like Spark.
  3. Productionized predictive models to forecast the effectiveness of advertising campaigns, including metrics like impressions, reach, clicks, conversions, and ROI.
  4. Building Scalable Simulation solution to model different inventory scenarios, including demand fluctuations, pricing strategies, and inventory allocation.
  5. Collaborate with cross-functional stakeholders from science team, product, engineering, operations, design, consumer research, etc., to productionize and deploy models at scale

Skills

Required

  • ML model deployment
  • inference infrastructure
  • low-latency systems
  • real-time systems
  • Spark
  • big data tools
  • predictive modeling
  • forecasting
  • simulation
  • inventory management
  • cross-functional collaboration

Nice to have

  • Lucene index
  • ad servers
  • bidders
  • yield optimizers
  • SSPs/DSPs
  • yield optimization
  • product recommendation
  • dynamic allocation
  • VAST
  • OpenRTB
  • legal compliance
  • ads regulations
  • CTV
  • ad tech

What the JD emphasized

  • low-latency real-time ad systems
  • large volumes of data
  • productionized predictive models

Other signals

  • ML models
  • forecasting
  • real-time ad systems
  • large volumes of data
  • productionized predictive models