We forecast cost-per-click in paid search by approximating latent competition through semantic, behavioral, and geographic signals derived from Google Ads auction logs.
Mar 13, 2026
We develop a multimodal hierarchical classification framework for cross-platform e-commerce product categorization, combining CLIP embeddings with late-fusion for state-of-the-art accuracy and a self-supervised recategorization pipeline for discovering fine-grained categories.
Dec 8, 2025
We develop neural product embeddings to predict sales of new products, addressing the challenge of forecasting with limited historical data.
Aug 1, 2025