Senior Applied Scientist, Agentic AI

Zillow

1d•$156,200 - $262,600•Remote

About The Position

The Senior Applied Scientist will work on Zillow’s Agentic AI team, which develops AI-powered experiences to help people navigate their real estate journey, understand options, make informed decisions and take action. The role works at the intersection of large language models and traditional machine learning, including ranking and personalization, and partners closely with engineering, product, design and science teams. The scientist will own complex and ambiguous applied-science problems from identifying opportunities and forming hypotheses through data analysis, modeling, evaluation, experimentation and production delivery. The role includes shaping which problems the team solves and how success is measured, developing and comparing machine-learning and LLM-based approaches, building dependable datasets and evaluations, and translating promising methods into reliable production capabilities. The position is remote; employees work from a physical location of their choice that must be identified to the Company. U.S. employees may live in any of the 50 states, with limited exceptions.

Requirements

Bachelor’s degree in Computer Science, Statistics, Mathematics, Electrical Engineering or a related quantitative field and 8+ years applying machine learning or advanced statistical modeling in industry, including production delivery; or an equivalent combination of education and experience.
Strong applied machine-learning and statistical foundations demonstrated through model development, evaluation and production impact.
Depth in a relevant area such as recommendations, ranking, search, personalization or conversational AI, with evidence of independently solving ambiguous problems and learning new domains.
Experience with large-scale datasets, experimentation or A/B testing, and model-performance measurement, including sound reasoning about data quality, bias, uncertainty and causal impact.
Practical understanding of LLM-based systems, including context, generation, evaluation and limitations, and the ability to apply that understanding to new problems.
Strong Python and SQL skills and the ability to turn analysis or a model into tested, reusable code that can be integrated into a production system.
Clear technical communication and a track record of delivering with engineering, product and science partners, with personal ownership and sound tradeoffs.

Nice To Haves

Experience with recommendation, ranking or personalization systems.
Deep hands-on experience building and evaluating LLM-based or agentic systems in production.
Experience with model distillation, supervised fine-tuning, preference optimization or small-model deployment, especially when balancing quality, latency and maintenance cost.
Experience evaluating multi-turn conversations, recommendation sets, session quality or longer-horizon customer outcomes.
Familiarity with large consumer marketplaces, sparse and delayed outcomes, cold-start experiences or real estate.
Research contributions demonstrating the ability to adapt new methods to practical problems and establish whether they help.

Responsibilities

Own ambiguous applied-science problems end to end, from identifying opportunities and forming hypotheses through data analysis, modeling, experimentation and production iteration.
Develop and compare machine-learning and LLM-based approaches to improve customer experiences, establish useful baselines, adapt relevant research and add complexity when evidence supports it.
Build reliable datasets and offline evaluations, and design online experiments that connect model quality to meaningful customer and business outcomes.
Partner with engineering, product, design and science teams to turn promising methods into reliable production capabilities while balancing quality, safety, latency and cost.
Write maintainable Python and SQL and contribute production-ready scientific components.
Share methods, findings and technical decisions with scientists and engineers.