Autonomous Drone
Constructed an automatic car capable of performing cruise, avoidance, and target tracking in a specific area.
Programmed on ARM-based microcontrollers in C to drive the engines, cameras, and various sensors
Applied traditional filtering methods for image segmentation and deep learning techniques for object recognition
Implemented fuzzy PID controller and optimized it with separate integral and well-tuned parameters
Achieved smooth beacon chasing with speed up to 3 m/s while flexibly avoiding crushing on obstacles
Troll Detector
Develop a chrome extension that detects trolls on Twitter by building sentiment analysis model and training it with the Sentiment140 dataset containing 1.6 million tweets from Kaggle
Applied advanced technical skills in data preprocessing, utilizing Google Colab for data cleaning, stemming, lemmatization, and stopword filtering
Explored several classification models like Bernoulli Naive Bayes, SVM, and LinearSVC, ultimately choosing lo- gistic regression for superior performance
Implemented the TF-IDF method for feature extraction and chose logistic regression for classification, achieving a model with 0.82 average precision and 0.83 recall
Route Troubleshooter
Chaired project of route traceback based on by-pass sniffing for SLA monitoring and network fault location.
Constructed spoofed source VIP packet sequences with increasing TTL fields to specific endpoint using Scapy.
Employed traffic mirroring service for packets capturing and asynchronous I/O with call-back filtering function.
Delivered on Advanced Anti-DDoS clusters while being elastically extendable via configuration files.