Built web-scale graph-processing solutions by implementing graph algorithms.
Built big data platforms twice. Performed problem modelling, data modelling and then several prototype-benchmark-migrate iterations for entire product suites.
Developed key frameworks, tools and modules implementing architectural patterns.
Directly contributed to several successful client implementations.
Led 5-10 member teams, provided design guidance to teams of large size (~100).

Published articles in LinkedIn on graph processing, security and stream processing.

Masters & Bachelors degrees in Computer Science and Engineering from Osmania and J.N.T. University respectively in India.

Big data : Hadoop (MR/Tez, Hive, HBase, Kafka), VoltDB, Netezza and Oracle
Cloud computing : AWS (EC2, S3, Cloudfront, Redshift), DYN - DNS
Languages : Java, SQL, Javascript, shell, Ruby, R.
Frameworks : Proprietary (FEBA, VIQ), Spring-boot, J2EE and RoR.
SDLC Tools : IntelliJ Idea, DBVisualizer, Erwin, git/svn, maven, jenkins, jira, confluence, SonarQube, puppet, docker.

Big Data Architect at Zeta Interactive Inc. (July 21 2016 - Present - Greater Boston Area)

At present, building multi-tenant data lake with HBase, Hive, Kafka, Hawq, Spark, Atlas and Spring-boot by prototyping and benchmarking batch/real-time data processes along with APIs, to enable interactive use cases, for a scale of ~100 Million rows per table per client.

Software Architect at Visual IQ Inc. (May 2012 - July 20 2016 - Greater Boston Area & December 2010 - January 2012 - Cochin, India)

Built web-scale graph processing solutions, with 1.3 Billion+ nodes and 10 Billion+ edges, for identifying ‘Connected Components’. Built 3 solutions as requirements, scale (100M to 1.3B+) and constraints changed:

Enhanced implementation of a classic algorithm for Percolation/Dynamic Connectivity problems- ‘Weighted Quick Union with Path Compression’ to a distributed variant embedded in a Hive UDAF.
Enhanced implementations of naive, block & diagonal GIMV(Generalized Iterative Matrix Vector algorithm) in MapReduce
An Iterative HBase implementation of GIMV variant in Hive

Led several successful proof-of-concepts with feature complete prototypes. Later migrated product suites (including data processes) to the new big data platforms.

POCs on Netezza, Vertica, Hadoop. Later migrate all products and processes from Oracle BI to Netezza
POCs on in-memory databases - VoltDB, memSQL, Cassandra. Later migrate backend of web interfaces from Netezza to VoltDB.
POCs on scalable, cost-effective data processing platforms with AWS Redshift, Hadoop.Migrate key batch processes from Netezza to Hadoop.

A Sqoop like data ingestion tool, from any database with JDBC interface to VoltDB. It is a multi-threaded, producer-consumer implementation that allows mapping tables from source to destination. Fine-grained control is enabled through SQL queries on source database and VoltDB stored procedures (invoked asynchronously). It moves 100s of Millions of records from Netezza to VoltDB at 16000 rows/sec rate.
A javascript container tag solution for processing site traffic to tag visits from untracked channels (ex: organic search, organic social, affiliates and custom domains). A de-facto component of all client implementations at Visual IQ (~80), it has no library dependencies and is less than 4KB uncompressed footprint.
Successfully refactored the production SaaS product suite to reduce technical debt, build frameworks and enable multi-tenancy,

Software Engineer at Altius Education, Inc (January 2012 - May 2012 - San Francisco)

Developed some of the key components in the course designer module - for the next generation online higher education LMS platform - Helix.
Setup CI platform with jenkins

Product Technical Lead at Infosys (April 2007 - December 2010 - Bengaluru, India)

An ORM framework for key-based operations on a given table. It implemented several patterns including Table Data Gateway, Optimistic Locking, Memento and DAO. It is employed as the de-facto interface for table operations in Infosys Finacle ebanking and mobile banking platforms used by some of the largest banks with highest online transaction volumes ex: ICICI, ANZ.
A DSL Type system framework for primitives and collections including a HashList. Some of the operations include sorting, deep-cloning and data format transformations ex: to JSON. It heavily used java generics and several key interfaces ex: Comparable. It is leveraged as a build-block by several other frameworks in Infosys Finacle ebanking and mobile banking platforms.
An Object Oriented RIA framework in javascript that fully encapsulates prototype.js and scriptaculous.js. It leveraged HTML5 data-* attributes, AJAX and Taglibs. It enabled portal-like interfaces with comprehensive widgets (ex: Transaction calendar), effects (ex: scaling) and features (ex: automatic client-side validations against types).

Consultant at Sierra Atlantic (November 2003 - April 2007 - Hyderabad, India)

Developed multi-threaded engine of an integration product between Agile PLM and JD Edwards ERP using Java, XML and XSLT. Later implemented it on Oracle SOA.
Developed complete web-interface from scratch for an EAI product suite with monitoring, RBAC & control features using Java, Servlets/JSP, HTML and Javascript

Certified programmer, developer, designer, architect
Certifications in Linux, VoltDB & Oracle 9i -SQL.
Coursera certifications in ‘Computing for Data Analysis’ (Johns Hopkins University), ‘Machine Learning’ (Stanford University) & marketing (University of Pennsylvania).

At Infosys - "Gem of the Quarter" award, for individual excellence.
At Sierra Atlantic - Contributed immensely in achieving 3 team awards.
At School - 5 year Scholarship from based on state level talent search examination.

Visa : H1B Visa valid till Jun 02 2019. Approved I-140 applied through Visual IQ
Hackathon Projects : R and RoR , GIS - Spatial Data.
R & RoR Demo : username:paul@datacustodian.org, password: Paul1234.