To gain employment with an organization/corporation of international repute where my technical and managerial skills can be used and honed effectively.
Overview
11
11
years of professional experience
Work History
Senior Software Engineer
Twitter
05.2018 - Current
Collaborated on various Data Processing Libraries like Scalding, Spark, Dataflow to run an End to End job in the most simplistic manner at Twitter.
Integrated Scalding/Spark/Dataflow jobs with Twitter Data Discovery and State-managed Services to run Scheduled jobs on top of Kubernetes/Aurora.
Build an Auto-tuning Service for Scalding Jobs to dynamically adjust the Containers Memory/Numbers necessary for the Job depending on the history of previous scheduled runs, resulting in greater Hadoop Cluster usage.
Lead a project to migrate existing on-premises infrastructure to Google Cloud Platform in order to perform batch jobs.
Collaborated with peers to develop an Event-Driven Scheduler for replicating data from various Data stores, including Hadoop distributed file system, Google Cloud Storage, Key Value NoSQL, BigQuery.
Develop a Managed Service to replicate data from BigQuery to our in-house Key-value store in order to serve machine learning models in the Serving path.
Consolidated all backend metrics into a single store that is used for visualization and to automatically optimize scala/spark jobs.
Tech Lead
InMobi
04.2012 - 05.2018
Implemented a new Spark Join API (Bucked By Bucket Join) that joins Map/Reduced Partitioned Datasets based on the available partitions.
It decreased overall duration of some crucial jobs by 80 %.This idea was inspired by Map Side join in Map reduce.
Build pipelines to measure how many people near the point of interest (Macdonald's, Universities) were seen in a certain time frame (hour, day, week) and within a certain radius (250m, 500m)
Collaborated on the creation of services that locate the user's location based on Lat Long, IP to calculate the Zip/City/State to efficiently target advertisements.
Built Geo-based location-hygiene pipelines to eliminate fraud Geo-inaccurate site IDs.