Distinguished Scientist

Zillow Zillow · Consumer · United States · Remote

Distinguished Scientist role focused on defining and driving the technical vision for long-horizon, supervised agentic systems at Zillow. This involves designing agent architectures, memory and supervision frameworks, and learning strategies, leveraging Zillow's real-estate datasets to train and adapt agents for high-quality recommendations. The role requires expertise in LLM post-training, evaluation methods, and architecting systems for persistent memory, tool use, and multi-agent collaboration in a complex, regulated domain.

What you'd actually do

  1. Own the end-to-end research and technical strategy for long-horizon agentic experiences across shopping, financing, and professional workflows, in close partnership with the Agentic AI, data platform, and product teams.
  2. Design and advance LLM post-training and evaluation methods (e.g., SFT, preference learning, RLHF/RLAIF, long-context modeling) tailored to supervised, high-stakes journeys in a complex, regulated domain.
  3. Architect systems that combine persistent memory, tool use, and multi-agent collaboration to deliver consistent, context-rich guidance over long timelines.
  4. Translate Zillow’s heterogeneous data (text, voice, behavioral, and structured real-estate/transaction data) into agent-ready knowledge and signals, in partnership with data and platform teams.
  5. Collaborate with product and design to define success metrics, evaluation frameworks, and experiment plans for agentic experiences, including human-in-the-loop supervision and safety reviews.

Skills

Required

  • PhD in Computer Science, Electrical Engineering, or related field, or equivalent experience
  • 10+ years of hands-on experience building and deploying large-scale AI systems
  • Several years focused on agent-based systems, multi-agent collaboration, or long-horizon conversational assistants
  • Deep, current expertise in generative and agentic AI
  • Expertise in multimodal foundation models, transformers, advanced reasoning models
  • Expertise in post-training techniques (SFT, DPO, RLHF/RLAIF, preference learning, etc.)
  • Track record of leading ambiguous, cross-functional initiatives from concept to production
  • Demonstrated impact in the research community through publications at top venues or widely used open-source contributions

Nice to have

  • AI planning

What the JD emphasized

  • long-horizon agentic systems
  • supervised agentic systems
  • complex, regulated domain
  • publications at top venues
  • long-horizon conversational assistants

Other signals

  • agentic systems
  • long-horizon agents
  • multi-agent orchestration
  • LLM post-training
  • evaluation frameworks