Working on building a large-scale distributed AI infrastructure, purpose-built for high-scale AI workloads (e.g., large-scale PyTorch and TensorFlow training jobs and inferencing models).

Worked on core components of Azure Service Fabric (microservices and containers orchestration platform) including developer experience, Reliable Services Programming model, Reliable Actor Framework, Replicated key-value store for stateful services, application hosting subsystem, container orchestration etc.

Worked on C++ AMP which is a compiler and programming model extension to C++ (introduced with C++ 11.0 in VS 2012) that enables the acceleration of C++ code on data parallel hardware like GPU. The data parallel computations performed on the device are expressed using high-level abstractions, such as multidimensional arrays, high level array manipulation functions, and multi-dimensional indexing operations. C++ AMP is also an open specification which can be implemented by anyone outside Microsoft on any platform.

Worked on developer experience for SQL Server data-tier applications in Visual Studio. It enabled a new way of developing, deploying and managing SQL Server data-tier applications by providing features like ability to import a live SQL Server database into the project, online/offline intellisense, seamless developer to DBA handoff by packaging T-SQL artifacts (like tables, stored procedures etc.) and deployment policies as a single immutable unit of deployment which can be deployed to a SQL Server (to create/update databases) or imported back into another SQL Server data-tier application project.