Kaushik Kumaran

Data Scientist
Walmart

Bio

Kaushik Kumaran is a Staff Data Scientist at Walmart Global Tech, with over six years of experience in retail data science. He currently works on the Pricing Data Science team, having previously contributed to Merchandise Assortment. With a background in companies like Mu Sigma and Lowe’s, Kaushik has consistently driven impactful GMV growth. A graduate of the University of Texas at Austin’s McCombs School of Business with a Master’s in Business Analytics, Kaushik also earned second place at the INFORMS 2024 poster competition, representing Walmart.

Optimizing Price Integrity: Walmart's Advanced Anomaly Detection System Using Quantile Regression, Neural Networks, and GenAI

Walmart’s Price Anomaly Detection system, currently deployed for both in-store and e-commerce items, uses advanced methods to identify pricing discrepancies. It moves beyond traditional classification techniques by leveraging Quantile Regression to define lower and upper price bounds, reducing the need for extensive labeling. The system focuses on predicting item prices, offering deeper insights into how various numerical inputs relate to actual prices. Neural Networks are employed to generate embeddings from product attribute features, predicting price bounds based on static categorical data. LightGBM enhances these predictions by incorporating historical prices, competitor data, and neural network outputs. To automate label predictions, Walmart uses GenAI, creating a feedback loop that improves model performance over time. This scalable, low-latency solution also ensures transparency through SHAP values, enabling model explainability.

Professional

Kaushik Kumaran

Data ScientistWalmart

Optimizing Price Integrity: Walmart's Advanced Anomaly Detection System Using Quantile Regression, Neural Networks, and GenAI

Get the Latest Updates

Data Scientist
Walmart