With the release of the OpenAI GPT Store, I decided to build a modern web application built with Next.js and Convex for AI-powered code review. This project demonstrates the integration of AI capabilities to assist in code review processes, providing automated analysis and suggestions for code improvements.

Designed and developed AI-integrated psychological assessment research at the University of North Texas, focusing on developing and deploying intelligent systems for cognitive and behavioral analysis. Engineered VR-based simulations for neuropsychological evaluation, processed multimodal data (e.g., EEG, eye-tracking, voice), and integrated Natural Language Processing (NLP) for automated scoring and insight generation. Collaborated on experimental design, statistical modeling, and human-subject studies involving real-time AI feedback and Human-Computer Interaction (HCI) optimization.

Next-Gen AI-Driven Neurophysiological Assessments Using Cloud-Streamed Virtual Environments Pioneered the integration of GPU-powered high-performance servers to stream immersive 3D virtual environments for real-time neurophysiological assessments. Leveraged Unreal Engine’s Pixel Streaming and Unity’s Render Streaming to deliver ecologically valid simulations directly via web browsers—eliminating the need for local high-end hardware and expanding clinical accessibility. Engineered a low-latency, high-fidelity system architecture capable of synchronizing biometric data (EEG, HRV, eye-tracking) with dynamic virtual tasks, ensuring scientific precision. Deployed scalable solutions on AWS, enabling secure, remote diagnostics and cognitive evaluations across diverse populations. This work underscores the transformative role of AI and cloud infrastructure in modernizing digital health and advancing brain-behavior research.

Built a machine learning pipeline to automate the detection of foliar diseases in apple trees, reducing reliance on time-consuming expert inspections and biological tests. Employed OpenCV (cv2) for preprocessing, including image resizing, grayscale conversion, and normalization to standardize input data. Implemented a custom Convolutional Neural Network (CNN) that achieved 93% classification accuracy on biotic and abiotic stressors affecting apple foliage. Benchmarked model performance against pretrained architectures—InceptionV3 (92%), VGG16 (62%), and VGG19 (63%)—highlighting the effectiveness of custom deep models for domain-specific disease detection. This AI-driven solution enhances early intervention capabilities and supports scalable, precision agriculture practices.