I am a Machine Learning Engineer at SOTAI, an AI startup where we focus on enhancing the transparency and interpretability of AI technologies. At SOTAI, we tackle the challenge of making AI decisions understandable and trustworthy.

Experience

2023 — NowMirascopeSoftware Engineer

2023 — Now

Python software engineer for LLM and mathematical machine learning toolkits.

Mirascope library:

Core library features, cookbook recipe examples, walkthrough markdowns

User experience review and feedback

Pytorch-Lattice:

implementation of base layers for model from scratch (lattice, random tiny lattices)

constraints checks for calibration and linear layers

creating/refactoring premade model classes for lattice/linear layers for code reusability and cleanliness

2020 — 2021Republic of Korea ArmyInfantry, Military Police

2020 — 2021

Seoul, South Korea

Squadron Leader for 6 months during mandatory military conscription for South Korea. Military Occupational Specialty: Munitions.

2017 — 2017SRI InternationalSoftware Engineer

2017 — 2017

Menlo Park, California, United States

Implemented two image processing functionalities for the software of a hyperspectral microscope.

Reversed a compound distortion stemming from the spectral "smile" as well as the pincushion distortion of the physical lens. Used classic image processing techniques such as gaussian blurring and edge enhancement to pick ideal calibration points, then mathematically reversed the distortions.

Designed an automated cell segmentation functionality for specific types of biological sample images. Using the average shapes of clearly individual cell samples, I generated custom convolutional filters for each color channel to achieve the results shown below:

2016 — 2016AhoraBack End Developer

2016 — 2016

Philadelphia, Pennsylvania, United States

Backend developer apprenticeship through Horizons School of Technology at Ahora, a Dutch startup aiming to connect the elderly with the best matching caretakers.

Working in JavaScript and MongoDB, I developed the database structure for both client and caretaker, as well as developed a matching algorithm based on factors such as availability, shift type, geolocation, and more.