Publications
Publications
Prompt Smart, Pay Less: Cost-Aware APO for Real-World Application
Developed a framework that reduces LLM inference costs by 30% while maintaining performance, directly addressing the primary barrier to enterprise AI adoption.
Jayesh Choudhari, Douglas McIlwraith, Piyush Kumar Singh, Snehal Nair
Recommender Systems Thesis
Reinvented Co-libry’s search strategy by replacing rigid, rule-based constraints with a first-of-its-kind intent-driven AI framework. By pioneering a novel model based on Empirical Fuzzy Membership, I bypassed the limitations of industry-standard filtering in high-sparsity markets. This shift moved the product from 'functional search' to 'discovery-led conversion,' delivering a scalable AI roadmap that aligns user behavior directly with bottom-line growth