Designed and implemented Big Data pipeline using Apache Spark, Amazon EMR, EC2, Redshift, RDS, MySQL to batch process metadata of millions of video ad airings.Applied data science, analysis (Jupyter Notebook, NumPy, Pandas, SciPy) and visualization (Matplotlib, Google charts, Seaborn) to identify and match video ad airing sequences and patterns.Designed and implemented ad airing sequencing algorithms, and sequence search (ElasticSearch).

Dec 2013 – April 2015: Data Science, Analytics, and Tooling EngagementDeveloped Dockerized full-text backend search component of IPython/Jupyter notebook JSON files for Knowledge Anyhow Workbench, a managed cloud environment easing the practice of data science (Python, ElasticSearch, Docker).Enabled IBM customers to gain insights from advertising and customer loyalty data by bootstrapping ETL pipelines, predictive models, named entity detectors, full-text search and interactive dashboards from their data using open source technology (IPython Notebook, scikit-learn, SOLR, UIMA, D3, MySQL).Mar 2011 – Nov 2013: IBM WatsonLed data ingestion efforts for IBM Watson commercialization.Improved runtime performance of Watson training time from one month to five days (Java, Lucene, Indri, Perl, triplet stores).Designed and prototyped three patent pending algorithms for Question/Answering systems.Implemented Hadoop MapReduce jobs to process Wikipedia and other large data sources and applied NLP algorithms to extract actionable knowledge (Java, UIMA annotators, ESG parser, Scrapy, BeautifulSoup).Applied data science to implement customizations to various IBM Watson stack components (runtime, NLP stack, ML stack – Weka, SPSS, web-based UI tooling, D3 visualization) to support new customer domains.

Mar 2010 – Feb 2011: IBM InfoSphere DiscoveryIdentified, designed and implemented integration of acquired company (Exeros) with IBM Information Server product in C#.Designed and implemented relationship discovery algorithms in Java. Contributed to iPhone application “BreadCrumbs” to scan 2D bar codes and track pharmaceutical products from supplier to customer.June 2006 - Apr 2010: IBM InfoSphere FastTrackDesigned and implemented Java J2EE-based metadata discovery and mapping services for “IBM InfoSphere FastTrack” built on WebSphere Application Server. Implemented Eclipse RCP UI.Designed, implemented, and patented transformation rule discovery data mining component.

Oct 2005 - June 2006 IBM Rational Data ArchitectImplemented Eclipse Rich Client Platform UI based mapping editor components and metadata discovery algorithms.Provided customer education, published IBM DeveloperWorks tutorials about SQL/XML query generation and relationship discovery.