List of my projects in AI, machine learning, and engineering
In-progress project: developing AI chat app for hospital group info, utilizing Language Model (LLM) tech, exploring LLM chaining, agent-based methods, and prompt engineering for accuracy.
Focused on long-term maintenance and improvement of Automatic Speech Recognition (ASR) model, updating with new datasets, training acoustic and language models iteratively.
Replicated SOTA paper on PCB defect classification, augmented dataset, fine-tuned models, enhancing results. Detailed results available in portfolio.
Developed AI system to detect LED defects on PCBs, adapting model to generate oriented bounding boxes for tilted LEDs. Established baseline and improved it.
Reproduced FrameExit paper's SOTA results in video recognition using HVU dataset, overcoming challenges with existing models, achieving 49.28% mAP.
Implemented novel Voice Activity Detection (VAD) algorithm using digital signal processing (DSP) techniques, reducing computational costs without deep learning models.
Implemented wake-up word feature for ASR applications, defining trigger phrases and experimenting with speech features like MFCC and PLP.
Developed preprocessing methods for speech signal enhancement: noise removal, dereverberation, and speaker separation, utilizing AI models for effectiveness.
A responsive web application for travel planning that my team developed in ICS 415.
Focused on long-term maintenance and improvement of Automatic Speech Recognition (ASR) model, updating with new datasets, training acoustic and language models iteratively.
Developing automated document information extraction system for handwritten and computer-generated text, employing segmentation, AI models, and OCR techniques.
Developed advanced auto-parts counter using object detection and tracking, aiming to replace weight-scaling method, showing promising results.