Jonathan Lobo

Drawing inspiration from Kaggle.com's March Machine Learning Mania competition, my friend and I used python and various machine learning classification algorithms in an attempt to predict the outcome of each game of the 2017 NCAA Men's Basketball Tournament. To view the source code and read our paper, follow the link below.

In a media environment saturated with deceiving news, the threat of fake and altered images in our lives has become increasingly apparent. The image forgery detection tool uses machine learning and computer vision techniques to detect copied, moved, or resampled sections of an image, to detect JPEG compression, and more.

The Internet of Things (IoT) refers to the worldwide network of billions of physical devices around the world that are now connected to the internet, collecting and sharing data. Internet of Things (IoT) security and privacy remain a major challenge due to the large scale, heterogeneity, and distributed nature of IoT networks. In this work, we discuss how the use of blockchains can not only enhance the security of IoT networks, but also enable new applications for IoT. We also conduct a feasibility study of the use of blockchains on constrained IoT devices.

World of Sweets is a digital board game inspired by Candy Land that is built in Java using JavaFX. The game has a responsive GUI and supports up to 4 players, including AI opponents.

Data-intensive computing and cloud computing are playing an increasingly important role in Internet services. Such services must tolerate failure, scale across a large number of machines, and support concurrent requests. This "Tiny Google" is a simple application that can index and search an arbitrarily sized collection of documents. The application is implemented both using Hadoop and manually using the MapReduce framework.

This project implements a functional simulator for a custom instruction set architecture based on MIPS. The simulator is capable of running programs written in the instruction set. In addition, an implementation of dynamic instruction scheduling using Tomasulo's algorithm is included which can provide statistics and a trace of any program.

The Keyword Automated Risk Assessment Tool, or KARAT, is a GUI-based java program that I developed by myself using Swing while interning at Springleaf Financial Services during the summer of 2015. The app, meant for in-office use only, allows loan underwriters to enter the key and date of any application the company has received. The program connects to the company database to pull data for that application and then scores it on the appropriate risk model. Finally, the program outputs a Microsoft Excel report with model-specific data and the results.

The OneMain Loan Decisioning App is a program that I co-developed using JavaFX in the summer of 2016 while interning at OneMain Financial. The program is a "game" of sorts intended to train branch employees in underwriting. Any company employee can register an account. Once logged in, the user is shown any number of de-identified credit bureaus and loan applications and is asked to score the simulated applications based on company guidelines. After doing so, the user is presented with aggregated data based on the reviews of others and is also presented with the system's actual decision. Written using Java, PHP, and CSS.

I contributed to the development of a new Safe Sleep Academy mobile app for the nonprofit organization Cribs for Kids, based in Pittsburgh, PA. The app was built for both iOS and Android using the Ionic framework.