Jul 2018 – Oct 2021 · 3 yrs 3 mos · Ann Arbor, MI

• Working on Criteo’s Retail Media Big Data Platform team building distributed, highly scalable data pipelines for processing and analyzing millions of e-commerce events. I am working on building Criteo’s unified self-service platform for Sponsored Products and Display Ads to help businesses increase product sales and build brand awareness. I worked on building Spark applications to perform complex data analytics such as Ad Attribution for getting key insights into marketing performance. Also worked on building a data pipeline for aggregating and storing data in Vertica and generating Ad campaign key performance metric reports for advertisers and retailers. Involved in cross team collaboration with Product and AX teams. Working on several technologies such as Hadoop, Hive, Spark, Presto, Mesos, Vertica.

Jun 2016 – Jul 2018 · 2 yrs 1 mo · Ann Arbor, MI

• Worked on Hooklogic’s (Acquired by Criteo) Big Data platform team for ingesting and processing terabytes of user behavior data. I worked on building a data pipeline for ingesting data, performing data transformations using AWS technologies, Spark, Hive and storing this data in Hadoop. I also worked on migration of legacy AWS based data processing applications to Criteo’s in house Hadoop cluster. This involved redesigning of existing applications & data model schema and building a brand new batch processing data pipeline.

May 2015 – May 2016 · 1 yr

• Worked in the Marketplace Seller Platforms team responsible for building a large scale data analytics platform for providing recommendations to Sellers on Amazon.
• Worked on 'BlastFurnace' which is a scalable data transformation framework for generating seller recommendations for restocking inventory, changing price of items, adding new items to their offering etc. The BlastFurnace framework is a collection of libraries for consuming data from various data sources, processing the data and persisting it to various data stores. The generated recommendations are shown to sellers on the Seller Central widget in order for them to take actions based on the recommendations.
• Worked on several AWS technologies such as SQS, Kinesis, DynamoDB, EC2, S3, EDX. Programming was done in Java.

Aug 2014 – May 2015 · 9 mos · Cleveland/Akron, Ohio Area

• Explorys provides a secure software platform enabling healthcare systems to aggregate, analyze, manage and research their big data to provide better patient care. Worked on the ‘Engines’ team, which is the core element of the Explorys Big Data Platform. My responsibilities involved writing complex map reduce jobs for analyzing a large volume of health care EMR data (millions of patients and billions of diagnoses, lab tests, medications, devices, outcomes etc.) and executing these jobs on the Hadoop and HBase cluster. Used SVN for code development and Jenkins for building and deploying.

Jun 2013 – Aug 2014 · 1 yr 2 mos · Austin, Texas Area

• Worked on Project ‘Solum’ which is an open source community driven PaaS. It is natively designed for OpenStack clouds and leverages numerous OpenStack projects, including Heat, Keystone, Nova, Trove, and more.
• The project aims at providing a platform comprising of multiple language runtime environments and backend services (just like Heroku add-ons) for running and managing cloud applications. It makes the process of converting source code into a managed application easier by providing git integration for submitting user code and also speeds the process using docker containers as the deployment unit.
• More information about Solum: https://wiki.openstack.org/wiki/Solum, http://solum.io/, irc channel: #solum on freenode.
• Project ‘RackerHunt’ at Rackspace Hackathon 2013.
• RackerHunt was a portal developed for helping the HR managers at Rackspace to recruit new candidates using Github. For techies, Github is the new Linkedin. Places like Github show what candidates have actually done instead of just reading about their potential. The idea was to pull Rackspace job openings data and find matching profiles on Github based on location, skills required and rank them. Profiles with higher contribution to Github repositories were ranked high and these matches were presented to the HR managers.

Jun 2012 – Aug 2012 · 2 mos · Columbus, Ohio Area

• Nationwide Enterprise Architecture
• Domain: BI & Analytics
• Installed and Configured IBM Rational System Architect for modeling Nationwide's business processes.
• Explored the features of Cognos Framework Manager and developed queries in Cognos Query Studio.
• Developed analytical reports using Cognos Reporting Studio for analyzing and gaining better insights from Nationwide's enterprise data.

Sep 2011 – May 2013 · 1 yr 8 mos · Columbus, Ohio Area

• 'Capacity Planning using Enterprise Architecture'.
• This research aimed at developing a greedy algorithm to automate the process of accurately predicting cloud resource requirements for applications in order to satisfy customer SLAs & for efficient resource utilization. This algorithm also made deployment predictions for the applications. This was my Master's thesis.
• Student Intern at The Ohio Supercomputer Center Jan 2012 - Mar 2012
• The project aimed at developing Thin Client based virtual desktop clouds in GENI.
• Developed 'Virtual Desktop Cloud Simulator' for providing Desktop-as-a-Service.
• A Real-Time, Model Driven Decision Support Mar 2012 - May 2012
• System for a Complex Service Enterprise
• The project aimed at building a tool for the City of Columbus to gain perspective into the as-is condition of its services in real time based on feedback obtained from real world sources.
• Developed a topic modeling, sentiment analysis and visualization system for analyzing 'letters to the editor' from Columbus Dispatch newspaper.
• Data Mining Project Aug 2012 - Nov 2012
• Implemented several data mining techniques such as preprocessing, classification, clustering, and association rule mining over Reuter’s dataset, which consisted of 21K news articles. This project was a part of the Data Mining course.
• 'OSU AcadNet' Mar 2012 - May 2012
• Developed an academic networking website to allow students of the CSE Department of OSU to interact with faculty members for their academic projects and research.

Aug 2009 – Jul 2011 · 1 yr 11 mos · Pune Area, India

• IBM PaaS Cloud
• I worked on Virtuoso which is a PaaS Cloud Computing System.
• Construction of virtual machine images from a collection of parts (software products, scripts, etc) and a base image.
• Topology model which defines the actual configuration of the application and the middleware to be executed on the cloud.
• Pattern development to deploy various applications on the Orion Cloud (private Eucalyptus cloud).
• Analysis of the ‘management fabric’ and implementation of support for handling ‘DBaaS (DB as a Service)’.
• LCAS (Log Collection and Analysis Service)
• The project aimed at collecting large numbers of log files (terabytes in size) generated by different cloud applications, storing them in Hadoop Distributed File System in Avro format and analysing them for extracting useful information.
• Map Reduce programs for various data format conversions.
• Creating Hadoop and Chukwa images and multi node cluster setup on Orion cloud along with setup scripts.
• Performance and storage analysis.
• Parser and Adapter implementation.
• Analysis of various failure scenarios of the cluster and provided solutions which can be used for failover support.
• Analysis of and tested dynamic scaling of cluster.
• Activities:
• Attended the Apache Hadoop India Summit 2011 organized by Yahoo India R&D at IISc Bangalore.
• Attended a seminar on Cloud Computing in Dec 2009 at Indian Merchants’ Chamber (IMC), Mumbai.

Jul 2008 – Jun 2009 · 11 mos · Pune Area, India

• The project aimed at developing a web service wrapper around an existing SunGard product “Abacus” which was a traditional web application. The purpose of the wrapper was to transform Abacus from a platform-specific web application to a platform-independent web service.
• Core areas of project are SOA (Service Oriented Architecture) and Web Services. Programming was done in Java.
• The project also included implementation of WS-Security (Transport Level Security and Message Level Security) and a C# .Net client to access the web service.