Minwoo (Michael) Kim

AI Researcher & Machine Learning Scientist

Career & Research

Scouter - AI Vision Analysis Assistant Platform

  • AI-powered Vision Analysis platform for Hearing impaired People
  • Real-time Zero-shot Object Analysis and Search
  • Real-time Super-Resolution and LLM usage with STT
SRGAN MobileSAM Asyncronous LLM Florence 2
View on GitHub

JobPT - AI Job Searching Assistant

  • AI-powered Matching Job Description to CV
  • Feedback on CV based on Job Description with numerical score
  • Personalized interview question generation and CV Revision
RAG ChromaDB Vision Parsing Knowledge Graph
View on GitHub

DINOv2 Backbone with TensorRT acceleration for Non-Destructive Inspection System Demo

Dinov2
  • Automated defect detection system using X-ray imagery
  • Implemented LBP and advanced image preprocessing
  • TensorRT optimization for real-time inference
DINOv2 TensorRT CUDA
View on GitHub

Chain of Thought (CoT) Agent Model Research

CoTAgent
  • Model Reasoning research like OpenAI's O1 model
  • Chain of Thought (CoT) based agent systems with self-aware iteration
  • Innovative sequential code execution approach
Reasoning Sequential LLM CUDA
View on GitHub

Closetmate - AI Fashion Assistant

Closetmate Project
  • AI-powered fashion recommendation system
  • Personal wardrobe management and styling suggestions
  • Image-based fashion item recognition
DeiT Vectorized Search K-means Clustering Boolean Masking
View on GitHub

J.A.R.V.I.S - Multimodal General Assistant

JARVIS
  • AI-powered Multimodal Assistant system
  • Understand Image, Text, Audio, Function Call
  • Response with Discord Bot
Function Call Image Understanding STT, TTS Reduce Latency
View on GitHub

Quantum Mechanics Research

  • Quantum Computing
  • Feynman Integral
  • Advanced Physics Research
Quantum Computing Physics Mathematical Analysis
View on GitHub

Mamba, Jamba, Hymba, Liquid Time-constant Network Research

  • Research New Architecture for LLM
  • Optimize Computing
  • Reduce Size of Model
Mamba Jamba Hymba LTC
View on GitHub

Paged Attention Research (KV Cache)

Paged Attention
  • Model optimization research in KV cache
  • Memory efficiency improvements
  • Performance optimization techniques
vLLM PyTorch CUDA
View on GitHub

Monosemantic Neuron Research

Monosemantic
  • Neuron specific Monosemanticity
  • Monosemanticity Checker with Sparse AutoEncoder
  • Understanding LLM black box
Sparse AutoEncoder Transformer Monosemantic Neuron
View on GitHub

Parametric Neural Networks Integration Dissertation

PNNI Project
  • Developed neural network-based approximation methods for Feynman integral calculations
  • Reduced computational costs significantly compared to traditional methods
  • Optimized for CERN LHC-like particle collision experiments
Deep Learning PyTorch Physics
View on GitHub

Reinforcement Learning with Monte Carlo Tree Search for Chess Game

RLChess
  • Optimized movement search with Monte Carlo Tree Search
  • Reinforcement Learning with Deep Q-Network
  • Parallel Computing for enhanced performance
MCTS DQN Parallel Processing
View on GitHub

Contact