Files changed (4)
+ \item Developed Hadoop application using Apache Pig for analyzing performance of A/B tests running on
+ the website. This application also served as a prototype for evaluating a possible transition from
+ \item Developed a distributed implementation of analytical model for recommending financial news and information based on client portfolios. Using Hadoop and Apache Pig, this new implementation was able to process 24 hours of news information for each client portfolio in under an hour, improving upon the original implementation which required over 12 hours to process a testing data set of 1/10 the size.
+ \item Created Jython based unit testing tool kit for Apache Pig. This provided functionality for automatically verifying the correctness of analytical models which reduced the overall defect rate. By using Jython to interface with Java based Hadoop and Pig APIs, this approach allowed developers more familiar with Python to quickly iterate on models in Pig via standard unit testing techniques.
+ \item Developed work flow system for Apache Pig which provided basic flow control and segmentation of individual models while still allowing for shared consumption of data extraction and cleansing pipeline.
+ \item Mentored junior developers on software engineering techniques for implementing analytical models and process automation.
+\item Research Developer -- Ackoff Center for Advancement of Systems Approaches, University of Pennsylvania 2006--2011
+ \item Developed framework for designing and conducting monte carlo simulations on software models created using in-house Agent Based Modeling (ABM) tool. Supports conducting both stochastic and monte carlo analysis. Also developed external API for easy integration with third party tools and models which reduced the time to completion for integration with third parties from weeks to days.
+ \item Implemented a data storage back end for retaining simulation state to be used in post run analysis. Previously, all data storage was done on a per model basis which resulted in duplicated work between many projects.
+ \item Developed a system for analysts to design models for prediction of abstract events and situations based upon the results of a completed simulation. These high level situations better express the outcome of a simulation than individual agent decisions making it easier to communicate the overall result of an experiment.
+ \item Leveraging the tools above, created a customized software stack for the Lockheed Martin Advanced Technology Lab as part of the DARPA sponsored ICEWS project (International Crisis Early Warning System). Due to the customizable nature components in the above tools, was able to develop an initial prototype in less than three days unlike the previous integration with the same group at Lockheed which took two months to produce the initial prototype.
+ \item Designed and implemented stored procedure scalability project. Substantially decreased the amount of system resources needed to execute the processes by performing query tuning and optimizations on SQL based stored procedures.
+ \item Audited and documented data model of in-house Warehouse Management System (WMS). Using this information, re-engineered data model and underlying database schema. Changes led to easier discovery and resolution of bugs which were causing inventory discrepancies.
+ \item Larvotto - Instant messenger bot which uses Markov chains seeded from log files of previous conversations to generate responses
+ \item PyVote - Python package for the modeling of legislative government bodies. Assisted in development and unit testing.
+\item Modeling and Simulation Methodologies--A Systems Approach to the Social Sciences, In: Modeling and Simulation Fundamentals: Theoretical Underpinnings and Practical Domains, Wiley Publishing, 227-270, Barry G. Silverman, Gnana K. Bharathy, Benjamin Nye, G. Jiyun Kim, Mark Roddy, and Mjumbe Poe
+For this to work you're gonna need some dependencies. On a Ubuntu system, you'll want these packages: