I’m Michelle D. Davies,

A Computer Engineer & Data Scientist.

My Portfolio

Browse my major projects, both finished and ongoing. Where there's a major project, there's a blog post to accompany it. Check them out!

I wrote a technical article during my 2021 IBM Technical Support co-op that troubleshoots setting up a Red Hat Enterprise Linux ("RHEL") 8 environment with the prerequisite libraries to use the IBM InfoSphere Information Server V11.7.1.2 product.

Link: https://www.ibm.com/support/pages/cant-find-operating-system-libraries-required-information-server-11712-installation

IBM Support Page for InfoSphere Information Server Install

My webpage: https://pages.github.coecis.cornell.edu/mdd94/mdd94.github.io/
Course Description: https://classes.cornell.edu/browse/roster/SP21/class/ECE/3400

Intelligent Physical Systems Portfolio

About 600,000 people in the United States, and 466 million people worldwide, are medically diagnosed as deaf or otherwise hard-of-hearing according to the World Health Organization (WHO). During COVID-19, many deaf/hard-of-hearing students are struggling to have their Individualized Education Plans (IEP) met in the wake of online learning. There are many solutions being developed to make it easier for able-bodied individuals to communicate with deaf individuals. However, few solutions exist to ease the barrier of communication for deaf people. So, I decided to develop a Voice to ASL and Morse Code robotic hand to decode verbal speech to visual signals and representations to bridge this communicative gap on the other end.

Voice to ASL & Morse Code

This Summer (2020), I did my internship at a startup called ZODAJ (http://www.zodaj-africa.com/#1) with students from Carnegie Mellon University and Stony Brook University to help develop part of a three-part system (Textbot, App, Website) to develop a contact-tracing system for COVID-19 in Senegal. I was on the website team. contributing to the development of the Hospital, Business and Government Portals.

Internship 2020 - ZODAJ

During my term as Web Director of WVBR-FM, I single-handedly re-designed & optimized the Cornell Media Guild Website for desktop, tablet and mobile. For this project, I utilized HTML, CSS, Javascript, and HTTP server connections, hosting server management and XML script.

Cornell Media Guild Website

This is a project that I particularly enjoyed because it ties my love of graphic design and art and colors with my love of circuits and robots. In this project, I display colors both in RGB LEDs and in terns of it's constituent R, G, and B values. This project has opened up opportunities for me to explore and analyze the random function, and system/computer colors as a whole. Check out the gallery and documentation to find out more about my final result! [UPDATE: I added a connection between this project and the My Image Suite Python script (see Projects > My Image Suite) by adding functionality so that I can display the RGB values of a selected photo's pixels through the board. The method for parsing the image is similar to this, with the added step of importing a photo in Arduino and the syntax for parsing the list of pixels].

The Mood Lights Project

To allow for more mobility for amputees we created a robotic hand that mimics the movement of the other hand. This will enable the user to do simple tasks like hold a cup or open a door. We noticed that many of the hands that were created did not have a way to save and replay their motions. So in the future we plan on making this function. Also buying a functional prosthetic arm is really expensive so with our project we are hoping that we will be able to make buying prosthetics more cost effective.

Mano816

During my high school years, I developed many different graphic designs for a variety of causes from Earth Day to Journalism, and I learned InDesign, Webflow, pure HTML/CSS/Javascript. Articulate and many more tools. This Project tab's photo album is dedicated to showcasing those projects from the past. You can also click the book next to the title to see additional projects from my time in the IB.

My Graphic & Web Design Exhibit

This was my final project for INFO 2950: Introduction to Data Science at Cornell University for Spring 2021.

Introduction Section (view Documentation for More):
One of my major ideas for my project was to look at the relationship between the **Economic Growth of a Nation** and the **Leading Causes of Death in a Nation** before COVID-19. Though morbid, I think death paints a fascinating insight into a country's socioeconomic status and provides enough scope to analyze the social constructs as well as the structural and institutional constructs that may exist in the context of each country, and how that affects who is dying the most from these major causes of death.

The main question that I have chosen to explore is: *What does a Nation's trends in Economic Growth tell us about its causes of Death?*
Within this umbrella question, the questions that I want to answer are:
* Of the six countries that reported the most deaths between 1999 and 2012 (China, The United Kingdom, Australia, Uzbekistan, Portugal, and Fiji), how has each nation's economic growth trend from 1999-2012? This range is due to the limitation that Economic Data only goes up to 2012, and the turn of the century brought about a wave of global economic growth and decline that had not been seen in centuries.
* What are the leading causes of death in each nation? What is the correlation between Economic Growth (% growth in GDP) and leading causes of death, if any?
* What demographics have been impacted the most by these leading causes of death? Is this a product of the countries' societal structures, or by chance?

Death and Economic Growth - Data Science Project

Michelle Davies & Shivali Halabe, ENGRC 3340, Fall 2020
Topic: What role do identity dynamics play in creating conditions for bias in learning algorithms? | Website: https://sites.google.com/view/identity-bias-in-ml-algos/home

We made a written publication that explores how identity dynamics foster bias in learning algorithms. Within this topic, we answered the following questions:
1. What makes a learning algorithm biased (datasets, assumptions, etc.)?
2. How do identity dynamics play a role in shaping these biases?
3. How is this narrative being reinforced by learning algorithms?
4. Based on the answer to the previous question, what are ways to avoid such biases?
5. What should we be teaching future algorithm developers and machine learning engineers to avoid discriminatory learning algorithms in the future?

Identity Bias in ML Algorithms

The Sound-Syncing Color LED lights is a system that takes the ambient sounds as an input, and outputs a visual, colorful representation of the given sound. I chose this for the Kinetic Art prompt because music and visual arts are two of the most pervasive art forms, and I wanted to apply Mechatronics and Signals Processing concepts to connect them.

Sound-Syncing Color LEDs

For one of my classes, we had to make a game that tests two players' reaction times. The way it works is that there are five rounds where, once each round starts, the first player to hit the button gets a point. The player that reaches 5 points first wins! But beware, if one player cheats by trying to press the button early, the other player gets the point for that round. Check out my gallery to see it in action!

Reaction Time Game

For this project, I created a prototype of an electronic piano with a basic ability to record all notes that have been played for the sake of songwriters that like to do improv. An upgrade that I was thinking of for this project was having the song notes auto format like a music sheet and save to a text or pdf file. Another idea for an upgrade that I had was to accommodate the shortage in button keys by having a switch that allows a user to scroll through notes and keys and reassign buttons to music notes of their choice. This is a prototype that I plan to upgrade in the future when I have the materials to do so. But for now, you can look at what I have at the moment on this page. Enjoy!

Arduino Piano-Songwriter

When I was in high school doing IB Higher Level Computer Science, we had to do an Internal Assessment project where we build a project for a client with a problem. I decided to build a simple photo-editing software for a friend of mine that was an IB Higher Level Art student, so that they count create their own photo-color filters in Obamacon photo style, but they can choose four colors to define the photo with. This project was done entirely in Python, from the photo-editing to the GUI that it is hosted in.

My Image Suite

Executive Summary
by Michelle D. Davies BSc ECE '22, Andrew Lin MEng ECE '21, and Kira Weinberg BSc MAE '22

Our vision for our project was to create a LoRaWan solution for TCAT to send their data from busses, to their central server. The challenge that we were addressing was to see if LoRaWan is a valid option for TCAT to move towards, and away from cellular which would save them money as well as help build the early open-source infrastructure for other companies to access affordable internet connectivity for their projects. For our project, We designed a prototype that could take its own GPS data, and send it through LoRaWAN and then map the gateway coverage of TCAT routes. This project proves the early potential for developing a LoRanWan system for communication for TCAT’s busses and other companies that might deploy very similar solutions. Once we collect the data, it is plotted on a map to see where the data was successfully sent, where the busses drove, and how fast the bus was moving at the time so that we can see where coverage exists. In our project, we have mounted one prototype to a bus that is able to drive any routes requested. This will help us to determine the effectiveness of our prototype for our application.

Our physical prototype consists of a 3d printed box that contains the devices that are mounted inside the bus’s electronics cabinets, is powered by the buses themselves, and is wired up to the top of the bus with an antenna shield/dome to protect the antenna from being damaged by the daily bus wash. While the device is in-field testing, we will find out if the shield can withstand the bus wash, and also if the antenna is able to transmit data through the shield or if it needs to be redesigned to be less limiting. The final vision of the mechanical box is one that is simpler to use, heat resistant, and able to protect the electronics inside and the antennae while minimizing their impact on their functionality.

The next steps in this project’s research and development would be to integrate the real information management system, Popufare, onto the device as well as connect it to the TCAT’s central server. From there, the web interface that displays the data could be hosted on a server in TCAT’s network so that the results can be updated in real-time and viewable to everyone with access to the internally-secured TCAT work network. The data could also be transmitted to the TCAT central server so that external applications such as Google Maps can access bus positions. We were only able to conduct field testing for half a day, but the system is wired up on the bus and is able to continue running, so this information will be available to whoever takes over the project in order to finish mapping out the coverage of local LoRaWAN gateways. From there, we would be able to see where there needs to be more coverage and potentially work with the city to see about adding supporting gateways to those areas, or change the prototype to have better connectivity. We will also learn about the various environmental factors that may affect our prototype’s function from geographical challenges such as cliffs, to weather.

Keywords: Lorawan, TCAT, Public Transportation, Internet of things

Internet of Things: TCAT Project

During COVID-19, one of the biggest questions households are asking is “how can we protect our homes from the spread of COVID-19?” One of the ways that COVID-19 can spread within an otherwise clean and isolated household is by touching items that were used outside, such as a used face mask for disposal reasons, and then immediately touching other household items before sanitizing their hands. Using a color sensor to Identify a mask, a light indicator to show the machine’s state, a robotic arm and hand with micro motor controlled movements, and a disposal mechanism, users can take their hands off their masks and limit their contact with the outside world.

Hands-Free Mask Disposal

This #BuiltLikeThat device is an Arduino-based project that os designed as a smart fitness tracker with clear, intuitive, physical signals that help users track their workout routines and goals from start to finish. This device’s primary goal is to bring the gym membership experience to your home while the ongoing pandemic occurs. The #BuiltLikeThat device accomplishes this goal by using a motion-tracker/sensor and a camera to detect a user’s position for set workout, and then using that information to count and keep track of each rep that the user does until they stop, For certain modes, such as the pacer test mode, the #BuiltLikeThat also comes with a timer to allow users to see statistics. Future versions will include a web interface that will be used to give users some analytics surrounding their usage of the device.

#BuiltLikeThat

During my term as Web Director of WVBR-FM, I single-handedly redesigned and updated the WVBR Website. The main coding I did for this project were HTTP scripting and Javascript to add even more functionality to this Wix website.

WVBR FM Website

For this project, I decided to get out of the "lights" realm for a bit and do something a little bit different, so I made a smart thermostat with a fan function where the fan will turn on at different speeds or off depending on the temperature recorded by the DHT Module in there. This was more of a challenge for me to put together in terms of the circuit itself and getting the right connections & code. You can check out my journey in doing this project in the gallery, and click the book next to the title to see the code as a non-text PDF!

Thermostat-Fan Machine

Inspired by a beginner's project from Girls Who Code, I decided to try to make a morse code machine that takes text input and outputs the corresponding more code signal for the entire phrase via the LED lights. There were many design considerations I made for this, like having dots and dashes appear on two different sets is lights to made them clear and distinguishable, thus improving readability. Overall, this project has been successful. When I have a greater budget and the necessary materials, I see myself upgrading this project by attempting to make a remote connection to the Arduino so that it does not have to be connected to work, and by mobilizing it so that it can "auto-deliver" any message to my family in these times of social distancing, Granted, I may need a sound component in case something happens to the circuit of lights or the message can't be interpreted for another reason.

Customizable Morse Code