Lead a team in Azure Core, managing platform for intelligently orchestrating the most impactful updates of Azure across millions of servers using Live Migration tech which seamlessly moves VMs across machines without down time. Introduced the concept of A/B experimentation of cloud platform changes for safety and velocity. This system understands Azure holistically with capacity, allocation, and customer aware co-ordination. Also lead another team which built the platform for external and internal Azure customers to control the Azure platform level changes seamlessly with 100% compliance for security. This platform manages 40% of Azure fleet with high availability and reliability and coverage across all geos and gov clouds. The most critical and sensitive customers of Azure are all using this platform to be successful in Azure.

Worked on designing and developing globally distributed platform for collecting information about Bing.com performance, errors and usage at client side (browser, mobile, windows). This system can handle 150,000 request per second and ~360 TB of data daily. Worked on optimizing performance of platform using .net core technologies to achieve high throughput on small 4 core machines and thereby reducing ~$3M annually on infrastructure cost.Designed and developed Near Real-Time big data streaming processing system for Bing.com which can process and merge 73 Billion logs (1,600 Tera Bytes of data) per day from various layers of Bing.com system and compute and aggregate metrics which is used for monitoring and improving Bing availability and quality. This system process entire data in memory to achieve end to end latency of ~1 minute. High fault tolerance and ~0% data loss was achieved using multiple layers of open source systems using Kafka and Storm

Created a highly distributed processing platform on top of Storm which scales to 12000+ processing nodes, which are synchronized using Zookeeper to achieve high availability. Highly efficient data compression and stream check pointing algorithms were developed for this system and patent was awarded for this ground-breaking idea which improved overall Bing availability and thereby saving millions of dollarsWorked on data curation pipeline which process and extracts data from various 3rd party data providers for Bing.com Local Search. This involves developing extraction models for ~80M data sets, improving the precision and recall of the models to achieve higher quality results for local search. Also worked on performance analysis pipeline which involves big data processing and analysis using Cosmos (Microsoft map reduce platform)

Worked on designing and developing Bing Shopping traffic and user analysis platform which mines user interaction patterns on Shopping search and improves the usability and thereby drive more traffic and revenue. This involves processing large volume of user interaction logs using cosmos and extract actionable signals which are fed back to the ranking and relevance of the results.

Worked as BI Developer for Bing.com team at Microsoft

Built end to end hospital operations management portal for Swedish Medical Center