Dylan Wong

Detail-oriented Computer Science student specializing in machine learning and artificial intelligence. Developed full-stack applications with RAG agents and implemented machine learning systems using Python, PyTorch, and TensorFlow. Conducted research on recurrent neural networks for digital signal synchronization, showcasing strong analytical skills and a solid foundation in algorithms, software engineering, and mathematics.

• Artificial Intelligence
• Intro to Machine Learning
• Introduction to Data Science
  and Big Data
• Python Programming
• Calculus 1,2,3, Linear Algebra, Statistics, and Discrete Mathematics.
• Data Structures
• Algorithms
• Software Processes
• Foundations of Software Engineering

RTX Sponsored ML Signal Processing

Role: ML Research Developer Tools: Python, TensorFlow, Keras, Jupyter Notebook, Matplotlib

• Developing a machine learning framework for digital signal processing under the guidance of Dr. David Detienne for an RTX-sponsored research initiative.
• Implemented recurrent neural networks to improve baud lock synchronization across noisy communication signals, and designed preprocessing pipelines to convert raw signal data into structured ML training datasets.
• Evaluating and optimizing RNN performance to improve synchronization accuracy and reliability in real-time signal environments.

Facial Deblurring with Emotion Preservation

  Role: ML Engineer Tools: Python, PyTorch, NumPy, OpenCV, Hugging Face

• Built a dual-model pipeline combining a CNN-based image deblurring network with a pretrained Hugging Face emotion classification model to restore blurred facial images while preserving emotional expression across a dataset of 70,000+ high-resolution images.
• Designed and compared four CNN architectures (encoder-decoder, shallow, shallow residual, and dilated), with the best model achieving a PSNR of 29.95, SSIM of 0.86, and an L1 loss of 0.021 — outperforming all other variants.
• Evaluated emotional fidelity of restored images using cosine similarity, L1, and L2 distance metrics against original emotion embeddings

Automated Stock Trader

  Role: Full-Stack Developer Tools: Python, JavaScript, React, Flask, Matplotlib, pandas

• Built a full-stack trading application delivering real-time updates on the FNGU and FNGD exchange-traded funds via web sockets.
• Developed a Flask backend to serve buy/sell signals, data visualizations, and user trading logs through RESTful API calls.
• Applied Simple Moving Average, MACD, and Bollinger Bands strategies to calculate and execute trades within defined time frames.