I’m a hands-on builder who is willing to take on anything that offers a chance to learn through experience.
I am curious about a wide range of technologies and enjoy continuously learning, experimenting, and expanding my interests across different fields.
Currently in progress and will be added soon.
An experimental pipeline for converting handheld indoor RGB video into rough 2D floor-plan drafts.
The project combines camera pose estimation, semantic segmentation, and monocular depth estimation to accumulate spatial evidence in a shared top-down coordinate system.
Highlights
- RGB-only indoor spatial reconstruction
- Semantic segmentation and monocular depth estimation
- COLMAP-based camera pose analysis
- Top-down free-space and wall-boundary reconstruction
- Experiment documentation, failure analysis, and feasibility validation
A reproducible data analysis and machine learning experiment for predicting whether an AI job belongs to the high-salary class.
The project compares five classification models, tracks experiments with MLflow, and manages the dataset with DVC.
Highlights
- AI job market salary data preprocessing and analysis
- Logistic Regression, Random Forest, XGBoost, LightGBM, and CatBoost
- Stratified train-test split and class-imbalance handling
- Accuracy, precision, recall, F1, ROC AUC, PR AUC, and log-loss evaluation
- MLflow experiment tracking and Model Registry integration
- DVC-based dataset version management
- Indoor spatial reconstruction from mobile video using depth estimation and camera pose analysis
- Data analysis and insight discovery through real-world datasets
- Reliable RAG pipelines and retrieval quality
- Infrastructure architecture, container orchestration, cloud deployment, and scalable system operations
