My name is Natalie (she/her) - welcome to my Github! I'm a sophomore studying computer science at Columbia University and a current AI fellow at Break Through Tech. Over recent months, I’ve been working in a team with Meta to develop a natural language processing model that detects demographic bias. 👀
🌱 I’ve been building my skills in linear algebra and applied machine learning in and out of the classroom. I’m currently working on a project under a professor at my school, Dr. Laine, to develop a convolutional neural network that conducts brain segmentation of 3D CTI images. The work and presentation for that is to be added soon!
Currently, I'm reading up for my upcoming work at the AlQuraishi Lab at Columbia University! As a machine learning researcher at the lab, I'll be investigating how cross-attention, masked language modeling, and distributed training can open doors for effectively predicting protein-protein interactions and binding affinity!
🤔 Feel free to take a look at some of my other projects I've conducted through the years:
-Fall AI Studio Project - Reddit Bias Detector: https://github.com/nataliecclaire/reddit_bias_sentiment_detector
-Break Through Tech Capstone - Predict Customer Subscription: https://github.com/nataliecclaire/customer_subscription_to_campaign_calls
-Linear Algebra Final Project - Cancer Classifier: https://github.com/nataliecclaire/cancer_classifier -> Here's the link to our final paper we submitted. Most of it is explaining the evaluation techniques we used and why they were beneficial: https://docs.google.com/document/d/1rZLcMH-zBnBLZ-WYvoDgVf1Mni-zYUYUnoLCB93gVeI/edit?usp=sharing
⚡Fun fact: I once coded an Arduino kit to create a Candy-Crush style game that looks like BMO from Adventure Time!
Given the limited RAM of the Arduino, my check_matches method could not be implemented without risking the RGB matrix lighting, so at least you can swap the candies around and sort them by color!
Thanks for stopping by!
Reach out to me through...
-Email: nataliecmckenzie@gmail.com
-Phone: (206) 371-8974