Applied Science Manager, Cba

Amazon Amazon · Big Tech · Seattle, WA · Machine Learning Science

Manager for a science team focused on developing transformer-based foundation models of customer behavior for forecasting and valuation. The role involves leading scientists and engineers in causal inference, sequence modeling, and experimentation, with a focus on methodological invention and translating research into production systems. The team builds shared infrastructure for understanding customer behavior across Amazon's retail business.

What you'd actually do

  1. lead Applied Scientists, Economists, and Data Scientists working at the intersection of causal inference, sequence modeling, and experimentation
  2. set multi-year research and product agendas in partnership with senior scientists, product owners, and business leaders
  3. work closely with a dedicated engineering team
  4. translate that work into production systems that move real launch decisions
  5. make learned representations estimation-aware

Skills

Required

  • machine learning
  • causal inference
  • sequence modeling
  • experimentation
  • ML management

Nice to have

  • deep learning
  • computer vision
  • econometrics

What the JD emphasized

  • 3+ years of scientists or machine learning engineers management experience
  • Knowledge of machine learning approaches and algorithms
  • 5+ years of building machine learning models or developing algorithms for business application experience
  • Experience building complex software systems, especially involving deep learning, machine learning and computer vision, that have been successfully delivered to customers

Other signals

  • developing transformer-based foundation models of customer behavior
  • learned directly from billions of behavioral events
  • shared infrastructure: one learned representation of customer behavior
  • causal inference, sequence modeling, and experimentation
  • methodological invention — making learned representations estimation-aware
  • closing the loop between experimental ground truth and model training
  • extrapolating short-horizon observations into year-ahead causal effects
  • translation of that work into production systems