Worked on building automation tools to test the worlds applications. Developed several types of models for the purposes of Computer Vision, OCR, and Reinforcement Learning related applications.

Working on tools and dashboards which help give insight on tracking documents. Developed an interface to enhance labeling for training future Machine Learning algorithms.

Worked on a Recruiting platform team. Developed several features, including a candidate recommendation system which used Machine Learning in order to compare directly the resume of the applicant to the job they were applying to. Also built system controls like permissions and used react to build the UI around setting those permissions.

Worked on a team using Machine Learning for automated testing. Helped with creating a tool which was capable of collecting, processing and storing webpages in order to be used for a classifier which could understand webpages.This tool uses a combination of base DOM and jquery in order to manipulate webpages and help label them for use in a classifier. The tool was deployed as a Chrome Extension and worked well across a wide variety of webpages.

Used supervised and Unsupervised learning techniques in order to solve problems in Online Social Networks and Online Crowdsourcing platforms.Detected fake and cloned profiles in Online Social Network using decision trees and also identified groups of users working together. The purpose of this work was to detect a Identity Clone Attack in order to evaluate legitimate friend requests.Detected fake or non-genuine reviewers in the Crowdsourcing platform Amazon Mechanical Turk, first by using semantic analysis through a technique known as lemmatization, followed by using a clustering algorithm to detect overlapping groups of malicious users.

Conducted research for efficient load balancing in Software Defined Networks using Network Function Virtualization. Implemented a framework to emulate Network Functions which was developed using C and the library Lpcap.Used techniques such as packet spoofing, packet modification on the line, and recalculation of checksums in order to fool the network into thinking modified packets were the original packets.The framework successfully emulated the traffic changing effects of Network Functions which helped verify the results of the research.This research was published in IEEE Transactions on Network and Service Management