Role Overview
Zebra's Data Science team delivers Professional Services for Enterprise SaaS Decision Management solutions that power multi-million-dollar supply chain operations. At the core is a Time Series Demand Forecasting engine driving downstream optimization for inventory, pricing, promotions, and workforce scheduling. As a Data Science Intern, you will work alongside senior data scientists and domain experts on real enterprise data—not toy datasets—benchmarking time series models and prototyping Generative AI and agentic workflows that redefine retail decisions.
Responsibilities
- Develop and benchmark demand forecasting models, from gradient boosters (XGBoost, LightGBM, CatBoost) to deep neural networks and foundation models (TiDE, TimesFM) and statistical baselines (ARIMA, Exponential Smoothing).
- Conduct data interrogation and EDA on complex retail datasets, detecting anomalies and root-causing demand shifts and model residuals.
- Power downstream decision solutions through mathematical optimization for allocation, pack optimization, and price elasticity modeling.
- Build anomaly and shrink detection models to identify phantom inventory and supply chain bottlenecks.
- Work with cloud data pipelines on Databricks and Azure, using PySpark, Delta Lake, and SQL.
- Implement model interpretability (SHAP) and build interactive dashboards with Plotly and Streamlit.
- Prototype GenAI and agentic AI workflows for automated data discovery, cold-start demand intelligence, and natural-language forecast validation.
Requirements
- Currently pursuing a Bachelor's, Master's, or Ph.D. (or senior-year undergraduate) in a quantitative field such as Computer Science, Data Science, Statistics, or Engineering.
- Hands-on experience with Python for data science and machine learning.
- Foundational knowledge of time series modeling and machine learning techniques.
- Working familiarity with SQL and cloud data platforms (Databricks, Azure).
- Strong analytical, problem-solving, and communication skills with genuine curiosity for exploring messy data.