Vijayan Shinde

Implemented Primal-Dual Learning Algorithm for dynamic pricing under inventory constraints for underutilized GPU inventory during non-peak hours at Vodafone. Simulated realistic, stochastic customer demand using logistic functions with randomness. Applied Machine learning models (LightGBM) for demand estimation and prediction. Achieved near-oracle revenues with regret consistently below 10% (theoretical threshold: 15%). Integrated primal-dual convex optimization to solve constrained pricing problems.

Data Analytics and Software Development

Engaged in Research, Analytics, and Mechanical engineering fields - Thermal analysis, Metallurgy, Finite Element analysis, and Manufacturing processes. Part of Designing a Quenching System for manufacturing structural mild steel rebar, using Finite Element Method to analyze transient heat transfer.