jadeddelta

May 2022 to July 2022

ML Research Assistant

I participated in Vassar College's URSI (Undergraduate Research Summer Institute) where I worked on a project to use Machine Learning to predict the outcome of where a user is looking on a screen given webcam data. In this project, I...

• devised a research and development plan with my team, which brought forward my first experiences with collaborative programming, Git version control, and GitHub's hosting platform.
• developed in Python, using packages such as TensorFlow for the ML model, and OpenCV for webcam data; along with using JavaScript with JsPsych for desgining the experiments that would be disseminated over Prolific.
• architected a CNN embedding network with LSTM processing, along with our own custom ETL pipelines for processing data with google's MediaPipe for facial landmark demarkation.
• presented our final model to the URSI symposium with a poster detailing our research and findings creating a model with a mean distance error of 18%.

May 2023 to August 2023

ML Research Assistant

For this summer, I was able to continue my work from the previous summer's URSI, working on the exact same project but on a different team of 2 other students. This time, I...

• expanded on computer vision machine learning research conducted last summer, implementing various models such as Siamese Networks and Attention Mechanisms.
• fabricated ETL pipelines to process 1000+ participants' data in a Python package, allowing for modular and scalable data processing of various facial landmark points.
• utilized services such as wandb for hyperparameter tuning, finalizing a Siamese CNN embedding model with attention and a custom triplet loss function.
• similarly to last year, presented our final model with a poster to the URSI symposium, with a final model that reported a normalized euclidean distance error of 7.18%.

October 2023 to June 2024

CogSci Programmer

Employed under Vassar's Cognitive Science department, I advanced forward a diverse array of projects, underneath the same professor that I was with during URSI. While working, I...

• refocused efforts on eyetracking machine learning research, redocumented data processing pipelines and standardized processing of 5000+ video clips using OpenCV and MediaPipe.
• explored new research literature, implementing novel solutions in computer vision such as visual transformers and novel loss functions in Tensorflow and Keras.
• assisted in maintaining jsPsych, writing TypeScript production code, implemented features for plugins that are unit tested, based on GitHub issues and discussions.
• spearhead a modernization of Vassar College's turing machine simulator, updating a 7-year-old website to Next.js with modern React and Redux practices.

September 2024 to Present

Assistant Research Software Engineer

After my work in the cognitive science department, I was selected to work under a POSE (Pathways to Enable Open-Source Ecosystems) grant, focusing my efforts primarily on <b>jsPsych</b>, where I currently...

• maintain and develop jsPsych as the most recently promoted core maintainer, engaging in code review, merging pull requests, and spearheading the work on new features.
• implement said features into new plugins and core functionality of the library, such as developing a plugin to determine if a user is wearing headphones or not for an experiment.
• collaborate with the jsPsych community, expanding community outreach and engagement within the repository and on the documentation, emphasizing a contributor-user experience.
• revamp the jsPsych documentation, written in MkDocs, creating new tutorials and updating both old plugins and tutorials to the newest versioning of jsPsych.