model-implementation

LLM-From-Abs-Scratch: A high-performance Large Language Model implementation built from the ground up in C++. Focuses on low-level optimization and architectural clarity.

PyTorch implementation of PaliGemma 2

Educational, from-scratch implementation of a LLaMA-style LLM using PyTorch to explore Transformer architecture fundamentals.

This repository is a collection of code, documentation, and other resources that support the management and automation of a Data Science project.

Implementing KANs with TensorFlow

This project utilizes machine learning to address the broad problem of spam through algorithms like Multinomial Naive Bayes and Logistic Regression; it can classify incoming emails as either spam or ham. This project aims to enhance email security and user experience while minimizing the risks of phishing attacks.

Detect sleep onset and wake from wrist-worn accelerometer data

This is the main project for 6th Kibo Robot Programming Round.

Predictive model for diagnosing diabetes in Pima Indians

A comprehensive machine learning web application for diabetes prediction, featuring real-time model evaluation and interactive performance visualization. This platform combines modern ML techniques with an intuitive interface to help analyze diabetes risk factors and make predictions.

Jupyter is a repository of various Jupyter notebooks that showcase data science and machine learning projects. Each notebook explores different datasets and techniques, providing in-depth analyses, visualizations, and model implementations.

Detect fraudulent transactions in financial data using machine learning algorithms.

paper implementation repo. i wish i come back to this :)

[2024~2025] MIMII 데이터셋을 활용한 이상탐지 + KSNVE AI 챌린지

End-to-end implementation of Faster R-CNN from scratch, including custom backbone, Region Proposal Network (RPN), and ROI head, trained without pretrained weights to study detection internals and training dynamics.

An ANN-based model was developed to predict the Position Sensor Offset Error (PSOE) in PMSM motors using a single input, accompanied by a user-friendly platform to enhance usability.

Predict which Tweets are about real disasters and which ones are not

Google Colab notebooks from my AI learning journey, covering machine learning, data science, and related topics.

A lightweight, low-latency proxy router for local LLM development. Routes requests to the best model for each task across multiple local backends with a unified OpenAI-compatible API. Features intent-based routing, memory-aware model management, streaming passthrough, and a hexagonal architecture ready for additional providers.