Mintae Lee

Software Engineer

Mintae Lee

Experience

Software Developer
SMBC Group: JRI America Inc., New York, NY

Dynamic and results-driven software developer with a proven track record of surpassing expectations and delivering exceptional performance. Throughout the years, showcased unparalleled commitment, adaptability, and leadership in complex technical initiatives

Demonstrated a keen understanding of cutting-edge technologies, DevOps practices, and Agile methodologies. Adept at steering projects, implementing technical solutions, and ensuring the successful execution of key initiatives.

Key Achievements:

  • Spearheaded the identification and resolution of critical components in internal application, implementing agile fixes and scalable solutions
  • Successfully assumed role as a lead developer for the team, implementing DevOps practices, ensuring continuous integration and delivery, and overseeing major releases.
  • Proactively engaged in Agile development methodologies and participated in critical AD services, ensuring seamless collaboration and adherence to ITIL practices.
  • Played a pivotal role in onboarding new developers, fostering collaboration through Agile methodologies, and leading weekly dev meetings to strategize and prioritize goals.

Education

Georgia Institute of Technology
M.S. Computer Science
Relevant Coursework: Machine Learning, Intro to Graduate Algorithms, Computer Vision, Software Development Process, Machine Learning for Trading
Cornell University
M.Eng. Computer Science
Relevant Coursework: Applied Machine Learning, Machine Learning Engineering, Data Science in the Wild, Virtual and Augmented Reality, BigCo Studio
New York University
B.S. Mathematics | B.S. Economics

Projects

The Terraformer
Unity, C#
Unity project that allows players to terraform a planet by harvesting natural resources to supply power to portfolio of transformative technologies
Buds - The Smart Menu
Javascript, Python, AWS
Cornell Tech BigCo Studio project that enhances dining experience by providing healthy, personalized recommendations
SMASH - VR Rage Room
Unity C#
VR sandbox space that allows you to relieve stress by recreating the experience of destroying objects in a virtual environment
Fridge Cataloging
Python, Tensorflow
Research Project that applied image segmentation and classification algorithms to catalogue items in refrigerator images

Contact

Please feel free to use the methods below to reach out to me for any questions or thoughts!

  • Twitter
  • GitHub

Contact

Terraformer

Game developed using Unity as a group project for Video Game Design class in GIT. The Terraformer showcases a dynamic gameplay environment where your mission is to harvest native resources.
Technical Features
  • Dynamic AI Scaling: Implemented AI techniques for dynamic difficulty scaling, ensuring a challenging and adaptive gameplay experience based on in-game challenges
  • Gameplay Mechanics: Developed and integrated essential gameplay mechanics, including object instantiation/destruction, interactions with enemy characters, and fluid character animations to elevate overall gaming experience.
Received Staff's Choice Award from the Georgia Tech's Spring 2022 Project Showcase

Buds - The Smart Menu

As the leader of the Buds team, I played a pivotal role in developing a cutting-edge digital menu aimed at revolutionizing the dine-in experience. Our project responds to the evolving demands of modern diners, leveraging technology to enhance menu accessibility, personalization, and nutritional information.
Technical Features
  • User-Centric Design: Led the creation of a user-friendly system allowing patrons to scan QR codes for personalized menu reconmmendations, allergen filtering, and calorie control.
  • Advanced Recommender System: Integrated tagging and recommendation algorithms to offer users tailored dining experiences based on preferences.
Selected as the top three project of the academic year from Cornell Tech's BigCo Studio

SMASH - VR Rage Room

VR group project developed using Unity, in which we created an immersive virtual reality (VR) experience focused on stress relief. "SMASH" recreates the cathartic experience of destroying objects in a virtual world, providing users a unique and guilt-free avenue for stress release.
Technical Features
  • Realistic Interaction Design: Developed a responsive system where the virtual weapon precisely follows users' motions, considering strength and speed.
  • Sensory Stimulation Integration: Implemented artificial senses to provide users with a tangible feel upon interacting with virtual objects.
  • Customization and Interactivity: Enabled users to customize tools, spawn destructible objects seamlessly, and dual-wield for an empowering experience.

Fridge Cataloging

Research paper focusing on leveraging image classification and segmentation to enhance kitchen organization, combat food waste, and facilitate informed decision-making regarding nutrition. The initiative aimed to develop a robust dataset of individual refrigerator ingredients and employed state-of-the-art neural networks for image classification and segmentation.
Motivation and Objectives
  • Addressing Kitchen Challenges: Recognizing the need for better kitchen organization and informed decision-making, the project sought to classify individual ingredients within a refrigerator.
  • Enhancing User Experience: By employing image classification, the intent was to assist chefs and home cooks in cataloging their fridge contents, suggesting recipes based on available ingredients, and minimizing food waste.
Datasets and Challenges
  • Curated Dataset: Curated a diverse collection of refrigerator items to overcome limitations in existing datasets. Incorporated images from multiple sources using techniques like web scraping and developed a final dataset with 63 classes and 70,189 images.
  • Annotation Efforts: Utilized the University of Oxford VGG Image Annotator to annotate a dataset of 76 refrigerator images to train neural networks for image segmentation and classification.
Methodology
  • Neural Network Exploration: Experimented with transfer learning using popular models such as ResNet-50, Inception ResNet V2, VGG16, and EfficientNet for image classification.
  • Image Segmentation Attempts: Initially explored non-neural network segmentation techniques, but due to limited success, transitioned to training a Mask RCNN model for both image segmentation and classification.
Conclusion
Despite challenges, the project showcased the potential of AI in kitchen organization through image classification and segmentation. The Mask RCNN model, in particular, demonstrated promising capabilities for cataloging refrigerator items, paving the way for future enhancements and innovations in this domain.