An Applied Scientist at Zillow operates at the intersection of machine learning research and software engineering. In this role, you are not just building theoretical models; you are designing, scaling, and deploying algorithms that directly power the most visited real estate website in the United States. Your work will directly impact how millions of people buy, sell, rent, and finance their homes.

The team's primary objective is to turn massive, unstructured real estate data into actionable, intelligent products. This includes driving the core algorithms behind the world-famous Zestimate, optimizing personalized recommendation engines, building sophisticated search ranking models, and developing computer vision models to analyze property photos. Because real estate transactions are high-stakes and deeply personal, the models you build must be both highly accurate and exceptionally robust.

As an Applied Scientist or Sr. Applied Scientist, you will collaborate closely with product managers, data engineers, and backend developers. You will be expected to own the lifecycle of your models—from exploratory data analysis and prototyping to production-level deployment and post-launch monitoring. It is a highly collaborative, intellectually challenging role where your scientific expertise directly translates into business value and user satisfaction.

The questions you will face during the Zillow interview process are designed to evaluate your theoretical foundations, your system-level thinking, and your ability to write clean, maintainable code. The following questions are compiled from real interview experiences of candidates who have gone through the loop. They represent patterns in what the hiring teams look for, rather than a list to be memorized.

Machine Learning & Deep Learning Foundations

This category evaluates your understanding of the mathematical and theoretical principles underlying modern machine learning models. You must be able to explain not just how a model works, but why you would choose it over another.

• Explain the mathematical difference between L1 and L2 regularization and how they affect model weights.
• How do transformer-based architectures handle sequential data differently than traditional Recurrent Neural Networks (RNNs)?

Preparing for an Applied Scientist interview at Zillow requires a balanced approach. You cannot rely solely on your theoretical knowledge, nor can you rely entirely on your coding speed. The successful candidate is a well-rounded practitioner who can seamlessly transition from mathematical theory to system architecture and clean execution.

Role-Related Knowledge – You must demonstrate a deep, intuitive understanding of both classical machine learning (regressions, tree-based models) and modern deep learning (transformers, embeddings). Be ready to explain the underlying mechanics of your models, including loss functions, optimization techniques, and evaluation metrics.

Problem-Solving & System Design – Zillow values scientists who can take an ambiguous product requirement and translate it into a concrete machine learning system. You will need to show that you can design robust pipelines, handle noisy real-world data, select appropriate metrics, and plan for model deployment and monitoring.

Execution & Coding – While some rounds may focus heavily on system architecture, you must be comfortable writing clean, modular code. Whether it is through a take-home project or a live coding session, you will be evaluated on your ability to manipulate data efficiently, implement algorithms correctly, and write reproducible code.

Collaboration & Communication – As an Applied Scientist, you will sit between engineering and product. You must be able to communicate complex scientific concepts to non-technical stakeholders and collaborate effectively with software engineers to integrate your models into production environments.

The interview process for an Applied Scientist at Zillow typically spans several weeks and is designed to thoroughly evaluate your technical depth, system design capabilities, and cultural fit. The process is rigorous but structured, ensuring that both you and the hiring team gain a clear understanding of the potential match.

The journey begins with an initial recruiter screen, followed by a technical screening phase which may include a hiring manager interview, a coding assessment, or a take-home project. If you pass this initial stage, you will move on to the comprehensive virtual onsite loop, consisting of multiple specialized rounds.

Throughout the process, Zillow interviewers look for candidates who demonstrate strong analytical ownership, a bias for action, and a collaborative mindset. They want to see how you think through problems from first principles rather than just hearing memorized solutions.

The visual timeline above outlines the typical progression of the Zillow hiring funnel. Candidates should expect to spend a significant portion of their prep time on the technical screen and onsite stages, as these represent the deepest evaluations of your coding, machine learning, and system design capabilities. While the initial stages focus on high-level fit and foundational knowledge, the onsite loop will test your ability to architect end-to-end solutions under realistic constraints.

To succeed in the Zillow interview loop, you must understand exactly what is being evaluated in each core technical area. The expectations are high, and the questions will probe the limits of your practical experience.

Machine Learning & Deep Learning Foundations

This evaluation area focuses on your theoretical grasp of machine learning. You must prove that you understand the mathematical foundations of the algorithms you deploy and that you do not treat machine learning models as black boxes.

Be ready to go over:

• Classical ML Algorithms – Deep understanding of linear regression, logistic regression, support vector machines, and ensemble methods (Random Forests, XGBoost, LightGBM).
• Deep Learning Architectures – Neural network fundamentals, backpropagation, optimization algorithms (Adam, SGD), and modern architectures like transformers and self-attention mechanisms.
• Model Evaluation – Choosing the right metric for the job (e.g., MAE, RMSE, MAPE, ROC-AUC, Precision-Recall) and understanding how to set up rigorous cross-validation schemes.
• Advanced concepts – Deep structural understanding of loss function modification, custom regularization techniques, and embedding generation for high-dimensional categorical data.

Example scenarios:

• "Walk me through how you would modify a standard loss function to ensure a model penalizes overestimation of house prices twice as much as underestimation."
• "Explain the self-attention mechanism in transformers and why it offers advantages over LSTMs for processing sequential property history data."

Applied ML System Design

Here, you are asked to design a production-grade machine learning system. You must show that you can think holistically about data collection, feature engineering, model training, latency, scalability, and feedback loops.

Be ready to go over:

• Data Pipelines & Feature Stores – How to ingest, clean, and store features efficiently. Handling missing data, spatial data, and temporal features.
• Scalability & Latency – Balancing model complexity with inference speed. Designing systems that can serve predictions to millions of users with sub-second latency.
• Monitoring & Maintenance – Detecting feature drift, concept drift, and setting up automated retraining pipelines.
• Advanced concepts – Multi-task learning, hybrid recommendation architectures, and designing real-time streaming feature pipelines.

Example scenarios:

• "Design an end-to-end system to recommend homes to users based on their search history, saved homes, and real-time click behavior."
• "How would you design a system to detect and flag duplicate or fraudulent rental listings in real-time as they are uploaded?"

Practical Coding & Take-Home Execution

In many pipelines, Zillow utilizes a take-home project or a practical coding round to evaluate how you work in a realistic environment. This is often a highly predictive stage of the interview process.

Be ready to go over:

• Exploratory Data Analysis (EDA) – Demonstrating your ability to find patterns, anomalies, and correlations in a raw dataset.
• Model Prototyping – Building, tuning, and evaluating a predictive model (such as a house sales value predictor) using standard Python libraries.
• Code Quality – Writing clean, modular, and documented Python code. Organizing your project logically and explaining your design choices.