Suryadeep

Data Scientist - Seller Experience @eBay

Toronto, ON, CA

Trained a multi-GPU Causal Action Guidance—a production-minded framework that recommends seller actions (e.g, free shipping, pricing tweaks) based on estimated causal effects, not correlations. CAG encodes a conversion DAG and learns a Bayesian structural causal model with posterior sampling to deliver predicted individual treatment effects (PITE) per listing, supporting both direct and total effects (do-calculus). In out-of-sample tests, our final large model (≈100k listings, for US) reached AUC ~79.5% and F1 ~59.5% within a <24h training budget; a causal what-if for free shipping showed conversion lift up to ~13% for eligible items. Collaborated cross-functionally with engineering, product, and trust teams to translate complex algorithms into intuitive user experiences and strategic platform capabilities.

St. Xavier's School, Doranda, Ranchi

Senior Secondary School, Mathematics, Physics, Chemistry & Computer Sciences

Lovely Professional University

Bachelor of Technology - BTech, Electrical, Electronics and Communications Engineering