Continuing to work as part of the same team as LinkedIn. Reorged into Microsoft. Developing the new generation Graph database. Currently leading efforts to develop a property graph-based secondary index, focusing on scalable bulk load data ingestion, query processing, and execution.

If you are interested in the product, please take a look: https://learn.microsoft.com/en-us/fabric/graph/overview

Working on Graph Algorithm and Analytics initiative for LinkedIn and Microsoft Azure.

• Working to bring suite of analytics and algorithmic capabilities such as path finding, community detection on really large scale graphs

Data Ingestion for Graph Database @ LinkedIn aka Liquid.

• Improved ingestion throughput by removing memory and CPU bottlenecks

• Developed quota and throttling mechanisms for various data verticals part of the Graph database

• Contributed to Kafka Opensource to introduce Java memory pools to improve garbage collection efficiency

• Improved performance and reliability of index bootstrap by eliminating the multi-region datacenter copies by 90%

• Served as TL for the data ingestion leading cross company initiatives to bring Graph technology to Microsoft

Data Infrastructure / Graph Database Team

I work on Liquid, an in-memory distributed graph database designed to serve declarative queries at high throughput and with low latency on LinkedIn’s Economic Graph. Liquid is a secondary global index consuming updates from different data verticals at LinkedIn in near realtime manner.

Liquid powers various social graph canvases ranging from basic 2nd and 3rd degree connections to things like finding the people who can refer you to a job you are looking at, finding the people who went to same school/company as you or who reacted to your post and many more.

I worked on AI Inferencing Platform, which enables intelligent scenarios across the Office 365 email, documents, and Sharepoint websites. I focussed for the platform's online infrastructure and NLP stack.

I contributed to the Office 365 Search team by enhancing machine learning infrastructure for email processing and document summarization. My work focused on real-time entity extraction and document summarization, which powered features like search autocomplete index, reply autosuggestions, and user personalization.

Moreover, I contributed towards development of generic the ML workflow engines, which empowered data scientists and ML engineers to ship represent their ML pipelines in production.

TA for 18-731 Network Security

http://users.ece.cmu.edu/~vsekar/Teaching/Spring18/18731/