Aug 2022 – Present · 3 yrs 10 mos · Bengaluru, Karnataka, India

• Azure Compute Org focusing on Platform services

Dec 2021 – Aug 2022 · 8 mos · Bengaluru, Karnataka, India

• Part of Cosmos Analytics Platform team. Primarily was responsible for multiple platform related fixes including Job Manager recovery, scheduler, input initialization and some optimization for fault tolerance.
• During this period I own the module which is used to help customers to analyze the cluster utilization by an xml file. Helped to integrate the generation of this file after migration of store to ADLS from cosmos. Also started an effort to mimic the entire cosmos end to end using VMs in local (effort was called Local Box).
• Led effort to optimize speculation - a feature of Job Manager which tries to improve task execution in YARN by scheduling duplicate execution on a candidate node which would guarantee a faster, better and efficient execution.

May 2020 – Dec 2021 · 1 yr 7 mos · Bengaluru, Karnataka, India

• Part of Cosmos Analytics Platform team (Externally offered as Azure Big Data Analytics). I work in a team which manages a component called Job manager which is responsible for running thousands of tasks on YARN cluster. During this role I was responsible for designing and shipping a tool which reduces the debugging overhead of the developers using a small file. The report was integrated into the job submission client which will generate the file after every job so that we will have this file for debugging if there comes any issue with the job.
• I was also responsible to ship an automated pipeline/bot which will auto analyze the issues filed by the customers and route it to the correct owning team based on the job error code. This bot helped us to route about 55% of the issues automatically to the correct time without manual intervention.