Overview
A 6-month Data Science internship in Bengaluru supporting Zebra's Enterprise SaaS Decision Management solutions, focused on time series demand forecasting and its downstream supply chain optimization applications.
What you'll do
- Research, build and evaluate time series forecasting models, from gradient boosting (XGBoost, LightGBM, CatBoost) and deep neural networks and foundation models (TiDE, DenseNet, TimesFM) to classical baselines (Exponential Smoothing, ARIMA) and Naïve models.
- Explore complex retail datasets through EDA, detect patterns and anomalies, and perform root-cause analysis on demand shifts or model residuals.
- Connect demand signals into mathematical optimization and heuristic engines for allocation, pack optimization, price elasticity modeling and workforce scheduling.
- Build unsupervised and semi-supervised anomaly detection models to identify phantom inventory, shrink patterns and supply chain bottlenecks.
- Work with cloud data pipelines on Databricks and Azure (PySpark, Delta Lake, Delta Live Tables, Unity Catalog, SQL) to clean, aggregate and prepare large-scale transactional POS and inventory data.
- Implement explainability techniques (e.g., SHAP, feature importance) and design interactive visualization dashboards (Plotly, Streamlit) to translate findings into business decisions.
- Prototype agentic and LLM-assisted tools for automated data discovery, schema mapping and ETL migration.
- Experiment with LLMs for cold-start demand intelligence, including analog matching, synthetic feature generation and semantic understanding when historical sales signals do not exist.
- Build generative co-pilots that synthesize promotional halo and cannibalization effects and generate natural-language forecast validation and root-cause narratives.
What you'll need
- Currently pursuing a Bachelor's, Master's or Ph.D. (or senior-year undergraduate) in Data Science, Computer Science, Operations Research, Statistics, Industrial Engineering, Applied Mathematics, or a related quantitative field.
- Solid understanding of statistics, machine learning principles and time series concepts (trend, seasonality, autocorrelation, stationarity, loss functions).
- Hands-on experience with pandas, numpy and scikit-learn.
- Solid command of SQL for data extraction, joining complex schemas, aggregation, window functions and data cleansing.
- A tenacious desire to dig into the "why" behind data anomalies, model performance variances and real-world supply chain behavior.
- A fast learner with the motivation and drive to rapidly upskill and absorb new tools, cloud environments and domain concepts.
Nice to have
- Exposure to Microsoft Azure and Databricks (PySpark, Delta Live Tables, Delta Sharing, Unity Catalog, MLflow).
- Experience with gradient boosting libraries (XGBoost, LightGBM, CatBoost) and deep learning frameworks (PyTorch or TensorFlow).
- Familiarity with or interest in modern time series architectures and foundation models (TiDE, DenseNet, TimesFM).
- Understanding of Operations Research and mathematical optimization (Linear/Integer Programming with PuLP, SciPy.optimize, OR-Tools or Gurobi).
- Familiarity with retail/CPG business metrics (sell-through rate, out-of-stock rates, price elasticity, GMROI).
- Interest or hands-on exposure to GenAI/LLM application development, prompt engineering or agentic frameworks (LangChain, LlamaIndex, AutoGen, CrewAI, Semantic Kernel, or OpenAI/Azure OpenAI APIs).
Details
- Location: Bengaluru, India.
- Work mode: hybrid, with adaptable hours.
Read the full description and apply on the company’s own careers page.