01 / Robotics & Computer Vision
Camera–LiDAR Perception with ROS 2
Contribution. Built a ROS 2 pipeline connecting YOLOv8n detections, camera–LiDAR fusion, vehicle-relative localization, and RViz visualization.
Evidence. Demo, technical report, and result bags with 1,305 timestamped output arrays at 0.25× replay.
Python · ROS 2 · OpenCV · YOLO
Demo & reproduction guide ↗
02 / Kaggle Competition · Team Project
Shopee Product Matching
Contribution. Contributed to a team retrieval solution combining Swin Transformer, NFNet, ArcFace, TF-IDF, and perceptual hashes.
Team result. Silver medal, top 4%. Includes the archived inference notebook and dependency notes.
PyTorch · Computer vision · Metric learning · Retrieval
Notebook & pipeline ↗
03 / Time-Series Forecasting
Stock & Energy Forecasting with PatchTST
Contribution. Applied the existing PatchTST model to Microsoft stock prices and AEP electricity demand using supervised training and self-supervised pretraining with fine-tuning.
Evaluation. Two univariate datasets, a seven-model baseline comparison, and accuracy/runtime summaries.
PyTorch · Univariate forecasting · Self-supervised learning
Code & recorded results ↗
04 / Reinforcement Learning · Course Project
Reinforcement Learning Strategy Evaluation
Contribution. Compared PPO, A2C, and DDPG in an NYU team course project using rolling training and validation windows.
Evaluation. Selected models by validation Sharpe ratio before evaluating the next trading period.
Python · Reinforcement learning · Time-series evaluation
Experiment & report ↗
05 / Healthcare & Statistical Modeling
Cervical Cancer Risk Modeling
Contribution. Collaborated on missing-data handling, exploratory analysis, and LASSO feature selection in R.
Evaluation. Compared models across sampling ratios using 858 patient records.
R · Statistical modeling · Healthcare data
Analysis & report ↗
06 / Natural Language Processing
Hate Speech Classification
Methods. Text preprocessing and BERT representations for hate speech classification.
Evaluation. Baseline model comparisons and ROC analysis.
NLP · BERT · Classification · Model evaluation
Explore the project ↗
