Working on new Data and AI Platform initiatives

Worked on system continuity for an external AML and Suspicious Activity Monitoring (SAM) product -- ended up working on a prototype and patent that used a graph database and news-feeds for alerting on current and historical suspicious activity patterns.

We have put into production around mid-2018 one of the largest -- if not the largest -- systems that Gartner calls HTAP (https://www.gartner.com/doc/2657815/hybrid-transactionanalytical-processing-foster-opportunities) and Forrester calls Translytical (https://www.forrester.com/report/The+Forrester+Wave+Translytical+Data+Platforms+Q4+2017/-/E-RES134282#). In the process, very excited to have taken Splice Machine on top of HBase and Spark behind one of the busiest application for disputes across the world -- I thank Monte for telling our story at the Keynote in HBasecon 2018.

Working on the data platform that can enable innovations happening in the payment industry.We have been using Apache HBase behind Apple, Android and Samsung Pay for sometime now!

Worked as the architect for the User-360 project that pulled user activity logs from online products to determine user actions across all Microsoft online offerings; in order to accomplish this goal -• Evaluated probabilistic user ID linking strategies across their various incarnations in different products;• Established a core set of dimensions to describe user context and activity and wrote data wrangling methods to conform data from 2 family of products (Bing Search and MSN) to a common vocabulary;• Experimented with a share-point based monitoring and intervention on data wrangling pipelines to maintain data quality;• Wrote user pathing methods on Cosmos (Microsoft big data infra-structure) to identify most prominent and likely user trails across products (user activity sequence).

Education, one of the oldest human activities, is going through a quiet but rapid evolution at Apollo Group, Inc. which runs the world’s largest “virtual university”, combining its subsidiaries, the University of Phoenix, and many others.Designed the Adaptive Learning Engine (ALE) for data-driven curriculum personalization,• Inspired by the company, Carnegie Learning, which has done wonderful data driven Bayesian work for teaching mathematics -- did a due diligence of Intellectual Property and helped in the acquisition of the company for struggling math learners.• Started the process of curricular breakdown and ontology creation into a semantic graph we referred to as the *Socratic Tree* -- this created a semantic graph for reasoning about learning content, assessments, and student mastery;• Created a process for designing "Tutored Formative Assessments (TFA)" -- these are assessments that collect student mastery signals on various parts of the Socratic Tree.• Rolled out Socratic trees and TFAs in pilot courses, and collected data using a Hadoop Platform that improved selection of learning content (as well as JIT help), assessments and their multiple choice answers, and reasoned about student remediation for mastery. The improvement was done through online experimentation, similar to A/B analysis by web companies.• Created semantic search appropriate for students in a certain course by scoring on relevance for the context the student is in, and the source of information.• Co-supervised Ph.D. work at Carnegie Mellon for helping find causal connections between actions during tutored problem solving in math and eventual mastery of subject.

Responsible for the Data Systems Architecture in APEX (Advertiser Publisher Exchange suite of products) for scalable and time-critical data collection, storage, retrieval, and processing through ETL pipelines on a data warehouse. The system is used for publisher revenue recognition and forecasting, advertiser campaign management, and web analytics. Led a team of engineers to develop Data Highway, a real time data collection platform, as well as the first real time data driven application for SEM (Search Engine Marketing) ROI analysis.

I designed, had the design reviewed and accepted by the Amazon Web Services management, and coded the first version of the Amazon SimpleDB service. This service is meant to lookup/discover entities based on their attributes.

Designed, coded, and enhanced an Enterprise Provisioning (bootstrapping) System -- a product named VSP 5000 -- for VoIP (Voice over IP) customer equipment.

I wrote algorithms for fast Computed Tomography (CAT) image reconstruction and artifact/defect removal in a new geometry of CAT scanners that involved electron beams.