Applied Scientist, Alexa-papi

Amazon Amazon · Big Tech · Sunnyvale, CA · Applied Science

Applied Scientist role focused on architecting and scaling next-generation memory intelligence and personalization capabilities for Alexa Plus, involving agentic AI, multi-modal signals, and large-scale deployment.

What you'd actually do

  1. Invent, design, and implement state-of-the-art solutions for previously unsolved problems in agentic AI, memory retrieval, and context management at scale.
  2. Build systems that learn from heterogeneous, multi-modal customer signals—conversations, device visual signals, ambient activities, purchase histories, and more—to deliver the right information, in the right format, at the right time.
  3. Design and deploy customer-feedback-driven, self-optimizing agentic architectures that serve millions of customers across Amazon surfaces, programs, and marketplaces.
  4. Contribute to a platform science team responsible for organizing and reasoning over diverse customer interactions to continuously improve personalization quality.
  5. Own the end-to-end science lifecycle—from hypothesis formulation and offline experimentation to online A/B testing and production accountability. Design evaluation that measure personalization quality and safety at scale, ensuring every launch is backed by statistically sound evidence.

Skills

Required

  • 2+ years of building models for business application experience
  • PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience
  • Experience programming in Java, C++, Python or related language
  • Demonstrated expertise in one or more of the following areas: Natural Language Processing (NLP); Recommendation, Search, or Advertisement systems; Conversational AI

Nice to have

  • Experience prioritizing and delivering projects on time in a fast-moving environment
  • Experience communicating complex information and solutions to senior stakeholders and influencing decisions
  • Hands-on experience with RAG architectures, LLM post-training, evaluation and reasoning pipelines, agentic system design, and scalable deployment.
  • Published research (peer-reviewed publications) preferred.
  • Strong familiarity with state-of-the-art Generative AI tools and frameworks (e.g., Claude Code, Langchain, Open Claw, or equivalent).

What the JD emphasized

  • previously unsolved problems
  • agentic AI
  • memory retrieval
  • context management at scale
  • multi-modal customer signals
  • self-optimizing agentic architectures
  • millions of customers
  • personalization quality
  • end-to-end science lifecycle
  • production accountability
  • personalization quality and safety at scale
  • statistically sound evidence
  • push the boundaries
  • generative AI
  • memory systems
  • agentic reasoning
  • production systems
  • millions of customers

Other signals

  • agentic AI
  • memory intelligence
  • personalization
  • multi-modal signals
  • scalable deployment