Michael Fatemi

Hey! I'm Michael Fatemi, a software developer and machine learning enthusiast. I have some experience with natural language processing and full-stack development, and I'm interested in biotech, robotics, aerospace, and quantum computing.

Education

University of Virginia
University of Virginia

2022 – 2026
BS Computer Science

Thomas Jefferson High School for Science and Technology
Thomas Jefferson High School for Science and Technology

2018 – 2022

Artificial Intelligence, Computer Vision, Web/Mobile Development, Data Structures, Macro/Microeconomics, Multivariable Calculus

Experience

Kyron Learning

December 2022 – Present; Natural Language Processing Intern

Developing natural language processing techniques to scale up the world's best teachers. Using large language models and reinforcement learning with human feedback.

Solar Car Team

August 2022 – Present; Embedded Systems

Working to develop a solar-powered car. Recently, I contributed to the development of a regenerative braking system to improve the car's battery efficiency.

Energy and Light Management Group

October 2022 – Present

Optimizing the training of neuroevolution potentials (NEPs) to predict properties of complex alloys. Using GPUMD and UVA's Rivanna cluster.

Internship with Dartmouth-Hitchcock Medical Center

June 2022 – Present; with the Levy Lab

Here, the goal is to use spatially-resolved transcriptomics to improve the diagnosis accuracy of cancer metastasis.
What I did specifically:

  • Analyzed spatial autocorrelation of colorectal cancer genomics data in Python with SpatialDE and SPARK-X
  • Developed convolutional neural networks and graph attention networks with Python/Pytorch to predict spatial expression of cancer-related genes with an AUROC of up to 0.89
Also developed some experience with Slurm/high performance computing. Work soon to be submitted to a conference.

Sound Camera Project

August 2021 – May 2022 (student research at TJ)

For student research at TJ, I made a "sound camera" with a Raspberry Pi, an embedded webcam, and a microphone array. This used a beamforming algorithm. If you want to read more about how I made it, see here!

Sound camera: Hexagonal microphone array with a camera on the top face and a Raspberry Pi below that.

Internship with Army Research Lab

June 2021 – August 2021; with Dr. Ei Brown

My primary research goal was to measure the efficiency of various laser crystals by detecting the amount and wavelengths of emitted light from each crystal.
What I did specifically:

  • Write Python programs to automatically detect the decay time of an emitted signal. This automation reduced analysis time by ~90%.
  • Create a real-time spectrum analyzer that enabled us to discover emissions peaks more efficiently, and decompose a single decay signal into various component decay signals

Project Caelus

November 2020 – November 2021

Project Caelus is aiming to be the first high school team to create a liquid-fueled rocket. I wrote C++ code to interface between Arduino and remote ground control software and integrate low-level sensor data in a read-control-actuate loop. I also helped optimize some printed circuit boards with KiCAD. The software was used to control valves during cold flow tests.

Volunteering

From April 2020 to April 2021, I volunteered at a computer science tutoring nonprofit called Codefy. I taught courses in Python and web development, and hosted workshops where students learned the fundamentals of machine learning, detected and classified traffic signs using computer vision, and developed games. I also served as a mentor, individually helping students learn Python, as well as leading courses in website development. I also developed an enrollment system that served thousands of students.

As a member of the student government, I rebuilt the main website with Typescript and React and built a website where students can share study guides. I also choreographed a hiphop dances for my year and organized other homecoming events.

Cool Things I've Made

Cactus Courseware


This is an open-source educational content website that my friends and I built. Educational content can be imported from GitHub and is updated automatically whenever contributors make changes. It includes animations and interactive content to cater to different learning styles. Uses React and Express as its frontend and backend frameworks, and MongoDB for data persistence.

StreetSweep

StreetSweep is an app that my friend and I created for a hackathon with the goal of helping policymakers and volunteer organizations by creating a heatmap of the prevalence and type of trash at each location. It uses class-agnostic object detection with a model adapted from a ResNet classifier to automatically detect garbage in photos of street litter.

WheelShare

An app to simplify carpooling: students can easily find people to carpool with for after-school events. 300+ users. Worked with school administration and sports coaches to get schoolwide adoption. Helps students who live far away from a school to have the same access to after-school activities as people who live nearby, especially if they are unable to drive themselves. Stack used: Frontend: Typescript React Backend: Node.js/Express, Prisma ORM, PostgreSQL database Hosting: DigitalOcean droplets, VPC for database security

EyeOS

An app that enables the disabled to control their computer mouse with their eyes: after a quick calibration step, the computer would detect which section of the screen they were looking at with 90% accuracy and move the mouse there. Then, via voice recognition provided by Google Cloud, they could click on those parts of the screen, type, search the web, and perform other tasks. Won 1st Place at the HooHacks 2020 Hackathon, an official hackathon organized by students at the University of Virginia. Utilized Haar cascades, DLib facial recognition, a custom iris detection algorithm made with OpenCV, and voice recognition from Google Cloud.

Other Things About Me

Rubik's Team 2019

TJ Rubik's Cube Club - President

This photo is from 2019. We had a lot of fast people in our year, and our team of eight won fastest in the nation for solving 25 Rubik's Cubes in 45 seconds!

Friends from track

Varsity Track - Middle Distance Co-Captain

Some people from the track team and I, after going to regionals in Spring of 2021.

Contact Me

michael [at] michaelfatemi.com

Or connect with me on LinkedIn