Projects

SAP Development Landscape Timeline web application

As an intern as BNSF Railway, I designed and created the full stack of a Flask web application that showed relationships between projects and events such as maintenance events or outages. All information was pulled from a SharePoint list using SharePoint's rest API and utilized recursive descent parsing along with SQLite databases to power it's rather fast backend. Working with server-less databases was a refreshing experience. The highly dynamic web app utilized Timeline.js to create an incredibly dynamic timeline with one of kind scaling and scrolling capabilities.

A dummy version (no real data) can be found here

ML model to detect instances of damage on train wheels

Working on the BNSF Railway Midas (Mechanical Image Driven Analytics System) under the guidance of Coleman Barkely I Developed computer vision deep learning models using the amazing Cogniac Corporation platform to detect instances of damage on portions of train wheels that existing models were struggling to identify. Being able to combine my passions for machine learning and computer vision was amazing.

Like many computer vision applications, my project started with understanding how the images were captured and processed might affect models created from the data. I then created python programs to analyze possible causes of detection imbalances from existing models and presented my findings. My data research lead to real conversation and change the image capturing system and my models continued training with hopes for deployment after the end of my internship.

Change Request Automation ServiceNow application

Using the ServiceNow platform, I collaborated with a fellow intern to create a web application that automated the creation and approval of emergency change requests within the SAP development landscape. This application would replace what was previously hundreds of emails each week, and decreased required user input fields by 38%.

Google Hash Code 2020

At the Google Hash Code 2020 Sharjah Hub, I put together a team of four fellow students, that collectively were from three different continents, and spoke five languages. I implemented a series of miniature scrum sprints in order to better structure our team’s efforts and also delegated tasks between team members as needed in order to mitigate issues that arose from conflicting approaches and preferences in programming languages. My 2020 team finished with 20,967,384 points in the 62 percentile of all 10,724 global competitors with two unique solutions I was very proud of.

United Arab Emirates Hackathon 2020

The United Arab Emirates Hackathon was a unique event for me, as it did not focus on programming solutions or other standard hackathon outcomes. Instead, the goal was to develop solutions and innovative ideas that would contribute to actual issues that the various communities of the UAE were facing under the competition motto: ‘Data for Happiness and Well-being’. I created a proposed solution to traffic congestion between emirates that utilized automated smart corridors and a Self-Adaptive Traffic Signal system based on real-time data. I also created a beta version of a Transit App feature that allowed users to plan routes within and between emirates using all available types of public transportation between destinations.

Helmet Identification – Visual Recognition AI Model

In order to demonstrate my Visual Recognition model, I used a real-world example of where AI could be implemented. On job sites, wearing helmets and similar safety gear is important for maintaining safe work environments. This model could inform a company when a worker is not wearing a helmet, or is, with a very high level of confidence despite a small set of unstructured training data. Below are sample screenshots of test inputs.

Bo the Chatbot

Bo the Chatbot, who hopefully you’re now familiar with, is powered by IBM Watson. Using Watson’s Discovery, Tone Analyzer, Assistant, and Natural Language Understanding for processing and analysis, Bo can answer most of the questions that reading through the entire site could. Bo will continue to grow and learn as the complexity of his algorithm increases. Give Bo the Chatbot a try at the bottom left corner of the site.

Image Processing - from scratch

As more autonomous vehicles hit the road and ‘Smart Drones’ take to the sky, image processing/recognition has never been more relevant. How does all of that work? Why should we care how that works when you can just import OpenCV into your project? Is this an efficient use of my time? I set out to answer only the first question. Using pixel by pixel processing and without using any image processing library, this program does edge analysis, object recognition, target blurring, and color analysis.

1 / 4

2 / 4

3 / 4

4 / 4

BattleBit

BattleBit is a game of Battle Ship programmed in C played on a local or remote server, where the game boards are unsigned 64 bit integers (8 by 8 board). Players can fire at each other, message one another, and print the board at any time.