• Major: Honors CS Student @ Georgia Tech
• Core Interests: ML Engineering • AI Research • Computer Vision
• Current Focus: Software Engineer Intern @ Cox Automotive, Undergraduate Researcher at Mind Brain, and Education (MBE) Lab @ Georgia Tech
• Portfolio: Visit My Full Portfolio →

Georgia Institute of Technology | Undergraduate Researcher
Lab: Mind, Brain, and Education (MBE) Lab | Advisor: Dr. Maysam Nezafati
Project: ASTaUND: Reproducible fNIRS Pipeline for Surgical Skill & Effective Connectivity Analysis
January 2026 – Present

• Objective: Building a standardized, BIDS-compliant Python pipeline for functional Near-Infrared Spectroscopy (fNIRS) to compare linear Granger Causality (GC) and symbolic Transfer Entropy (TE) as measures of directed prefrontal–motor connectivity, and to develop delay Symbolic Transfer Entropy (dSTE) for quantifying nonlinear information flow related to surgical expertise.
• Signal Processing: Implemented a Homer3-aligned denoising stack—Recursive PCA (rPCA) motion correction and Short-Channel Regression (OLS)—in MNE-Python to isolate cortical hemodynamics from systemic noise across the ASTaUND Expert–Novice cohort.
• Linear Connectivity: Developed a frequency-domain bivariate GC pipeline (VAR-based) on ROI HbO time series as a linear effective-connectivity baseline for Expert vs Novice contrasts.
• Symbolic TE / dSTE: Implemented and extended delay Symbolic Phase Transfer Entropy (dSPTE) (Neural-Gas symbolization, multi-lag discrete TE, multivariate conditioning), then generalized the framework toward amplitude-domain dSTE—with continuous TE validation via IDTxl—as a nonlinear counterpart to GC.
• Technical Stack: Python (NumPy, SciPy, Pandas, statsmodels), MNE-NIRS, IDTxl, YAML Configuration Management, BIDS/SNIRF Standardization, and modular stage-based reproducibility.

Category	Tools & Technologies
Languages
Frameworks
AI/Data
Tools

SmartPath AI > An adaptive educational platform that uses RAG and Neo4j knowledge graphs to transform static PDFs into interactive, personalized study paths. > Python LangChain OpenAI RAG Pipelines

AI-Powered Prosthetic Arm > A bionic control system that uses unsupervised autoencoders and computer vision to translate raw EMG muscle signals into natural, 6-degree-of-freedom hand gestures. > Tensorflow Unsupervised Learning Python OpenCV MediaPipe

Recruit Ready > A real-time interview practice software with AI agent-powered feedback on body language and speech. > Python Mediapipe OpenCV Google Gemini 2.0 FastAPI

Brain-Computer Interface System > A Brain-Computer Interface (BCI) system that translates SSVEP EEG signals into real-time mouse cursor commands using frequency-domain machine learning. > Scikit-learn LDA SVM Filter Bank CCA Python SciPy

PreVizAI > An AI-powered pre-appointment patient intake system that streamlines medical check-ins with voice-first EMR integration. > Python Flask API OpenAI Whisper Cedar OS SDK

• Portfolio
• LinkedIn
• Email: vinh.pham2529@gmail.com