• Designed and developed Google Cloud Trace Exporters for C++ OpenTelemetry client using Cloud Trace V2 API to get near real-time performance insights on VMs, containers, App Engine projects or other applications.• Instrumented, dockerized and deployed gRPC-supported microservices to Kubernetes cluster on the Google Kubernetes Engine to analyze performance of OpenTelemetry Instrumentation API.• Built a multithreaded Batch Span Processor for C++ OpenTelemetry SDK for batching and conversion of spans to exportable representation to increase the efficiency of the export process to the configured backend vendor-specific exporter.• Added forking support to Span Processors and Exporters to extend the reach of OpenTelemetry to a vast array of applications.• Languages/Tools: C++, gRPC, Docker, Kubernetes, Google Test, Google Benchmark, Google Mock, Google Protocol Buffers, Bazel.

• Worked with the WhatsApp Data Infra Team• Designed and implemented Dataswarm pipelines and Erlang modules to enable a built-in timeline-tracking functionality for WhatsApp telemetry logging stack and perform version-by-version comparison on real-time data analytics platform such as presto, hive, scuba, etc.• Developed building blocks of a Configuration-to-Data (C2D) reliability system to monitor, analyze and measure reliability of WhatsApp Experimentation platforms.• Languages/Tools: Python (Dataswarm), Erlang, SQL (Presto Query Engine + Hive), Scuba, Git, Mercurial

Designed a framework for Time-Series abstraction for HealthCare Analytics in collaboration with Weill Cornell Medical College.Built a clustering Machine Learning model for pattern-analysis of sleep data and prediction of sleep quality.Created a model to learn similarity functions from patients’ data by utilizing Metric Learning.

Developed the classic game "Reversi" on various topological spaces such as Moebius strip, Klein Bottle etc. with 3-D animations and implemented VR renders for the same.Rendered “AI/Human vs. AI/Human” capability for the game using Minimax and Alpha-Beta Pruning artificial intelligence algorithms with naïve heuristic functions.