Working on content delivery, personalization, and next-gen viewing experiences on Prime Video. Focussed on Ownership, Delivering results and Collaboration.

• Led Prime Video - Alexa service meeting 300 ms latency SLA and generating conversational voice text streaming LLM summarization processing 100K queries daily.

• Architected distributed microservices for ML inference, achieving 99.9% availability under high-concurrency production traffic.

• Optimized large-scale PyTorch inference pipelines, reducing P99 latency 3x through parallelization and memory-efficient caching.

• Designed feature gating framework to support controlled rollouts, A/B testing, repeatable evaluation of 10+ AI features

• Established REST API contracts, unit testing, and CI validation pipelines reducing regression defects 30% before release.

• Identified system bottlenecks and improved performance through scalable caching and service orchestration.

Developed reasoning framework to enhance semi-structured & multi-modal Table QA.

CSE 575: Statistical Machine Learning, Prof. Kookjin Lee

CIS 565: AI in Business, Prof. Julian Lehmann

CIS 508: Machine Learning in business, Prof. Asim Roy

Built predictive models for demand forecasting cutting supply chain costs by $4M.

Designed big data pipelines & data models, that manage 1TB+ data in Azure Data lake and S3 buckets

Built predictive models from employee KPI data, improving performance rating accuracy by 25% enabling data-driven evaluations.