Grace (Unnseo) Park

Hello, I'm Grace Park .

A

I am currently pursuing a Mater's degree in Computer Science at Carnegie Mellon University's School of Computer Science, building on my undergraduate studies in Artificial Intelligence, also from CMU. I will be graduating in May 2025 and am actively seeking full-time opportunities starting after my graduation.

I am a Software Engineer with a strong passion in applying technology and Artificial Intelligence to create a positive change. I am especially interested in the healthcare and medical industry. I believe that AI and Machine Learning deliver valuable insight in healthcare due to its ability to accurately process large amounts of data beyond the scope of human capability.

Experience

CMU Data Interaction Group

Undergraduate Researcher

Conducted research on the application of reinforcement learning and transformer models for Sepsis Treatment Prediction.
First Author of "How Consistent are Clinicians? Evaluating the Predictability of Sepsis Disease Progression with Dynamics Models" presented at ICLR 2024 Workshop TS4H. (Link to Paper) (Link to Poster)
Designed and trained transformer-based dynamics models using Python and Pytorch, experimenting with varying hyperparameters to optimize model performance and fine-tune large models for improved accuracy.
Developed and implemented experiments using dynamics models to evaluate the feasibility of predicting disease severity changes based on clinician actions using reinforcement learning with the MIMIC-IV dataset.
Applied advanced data preprocessing techniques to impute missing data and prepare the dataset for modeling.
Skills: Python, PyTorch, Scikit-learn, Google Cloud Platform, Numpy, Matplotlib

January 2023 - Current | Pittsburgh, PA

CryptoLab

Research Engineer Intern

Implement and compare the performance and speed of different optimizers such as SGD, Adam, and Adagrad for regression models of homomorphically encrypted data.
Contribute machine learning features to the HEaaN library.
Improved the training speed of logistic regression models by 3.6% and achieved 95% accuracy.
Skills: Python, PyTorch

July 2022 - August 2022 | Seoul, South Korea

CMU AI Makerspace

Undergraduate Research Assistant

Set up various robots such as Misty and Kinova robotic arm in the AI makerspace and prepare for opening.
Construct manuals and descriptions on how to use each robot.

September 2021 - May 2022 | Pittsburgh, PA

Mediazen

Software Engineer Intern

Assist in the development of computer vision technology including gaze tracking and lip reading.
Tools: Python, OpenCV

June 2021 - August 2021 | Seoul, South Korea

Projects

                
Anyways...
                  An AI created to help you stay focused on your discussion 
                  by saying “anyways… let's get back to our topic".
                
AccomplishmentsTools: PyAudio, spaCy library
Uses real time speech-to-text technology.
Keyword similarity analysis from the spaCy natural language processing library.
Designed an algorithm for detecting off-topic sentences using keyword similarity.

Campus Map
                  A prototype android app to help users find the shortest route between two locations on campus. 
                
AccomplishmentsTools: Google Maps API, Kotlin
The app measures the distance the user walks to reach the destination from the starting point.
The leaderboard displays the shortest paths that other users took.

Drive! Game
                  A car-driving game that has a top-down view.
                
AccomplishmentsMy first project!
Tools: Python, 112-graphics
The player's car is always at the center and the background mimics 
                    the real-world roads with obstacles that the car must not collide into, 
                    such as other cars, bicycles, etc. A map generation algorithm creates 
                    new random maps every time. 
                  
Robot BlackJack
                  Blackjack! Interaction with a Cheating Robot During Complex Gameplay.
                
Accomplishments
                A replication study of “No fair!! An interaction with a cheating robot.” by Short et al.
                It was found that in a simple game of “Rock, paper, scissors" between a robot and a human participant, 
                the participant will be more socially engaged and perceive the robot as acting more agentic if it cheats. 
                We examine whether these differences between cheating and non-cheating behaviors remain under a more 
                complex game “Blackjack."

Leadership

                
Korean Student Association
                  Board member of CMU's KSA
                
AccomplishmentsI was actively involved in planning for various events such as food sales, mentor programs, 
                    Korean Independence Day events on campus to promote Korean culture.
I served as Social Media Chair for KSA. I created and managed the official Instagram account and regularly posted stories and posts.

                Sept 2021 - Current | Pittsburgh, PA
              
Tartan Ambassador
                  Tour guide for Carnegie Mellon Univeristy's campus tour. 
                
AccomplishmentsI led hour-long in-person tours across campus for a group of up to 20 visitors which include prospective students and families.
I learned a wide variety of information about CMU to expertly support visitors and articulate Carnegie Mellon University's brand story.
I answered any questions and talked about my personal experience at CMU.

                May 2023 - Current | Pittsburgh, PA
              
Beyond Coding
                  Teacher of "Introduction to Programming" for middle school students. 
                
                Accomplishments
                I taught beginner to intermediate-level programming in python to a group of 
                    10 middle school students.
I prepared for weekly 3-hour long sessions of lectures and activities.
I assisted students with their individual projects.
                  
                Jan 2021 - Aug 2021 | Pittsburgh, PA