Expertise in designing, training, and deploying AI models using supervised and unsupervised learning, with proficiency in PyTorch, TensorFlow, and Keras.
Specialized in Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), fine-tuning, text embeddings, semantic search, and prompt engineering.
Proficient in object detection, image processing, and developing real-time computer vision systems using architectures like YOLO.
Skilled in data analysis, preprocessing, and building end-to-end machine learning pipelines using Python, SQL, NumPy, and Pandas.
Engineered a deep learning-based web security classification system utilizing CNN and RNN architectures to accurately detect and isolate malicious web request payloads. Developed custom word embedding techniques to preprocess raw HTTP request text streams.
Architected an advanced RAG application combining vector embeddings and semantic search strategies to enable conversational document analysis across complex uploaded texts. Integrated document chunking pipelines and text embedding models to map and retrieve relevant document indices.
Developed a real-time computer vision system utilizing the YOLO architecture to identify and interpret traffic light states (red, yellow, green) within autonomous vehicle simulation contexts. Designed data transformation pipelines to handle diverse lighting and angle constraints.
Built an end-to-end machine learning pipeline to identify fraudulent credit card transactions within highly imbalanced operational datasets. Implemented rigorous data preprocessing, scaling, and class-balancing strategies to address severe anomalies in transactional records.
Provided on-site technical support and resolved PC hardware/software issues as an IT Support Specialist at SVA.
Engineered a deep learning-based web security classification system, achieving 92% accuracy and an A+ grade for AI Graduation Project.
Architected an advanced RAG application, reducing context retrieval latency to 400 ms.
Developed a real-time computer vision system for traffic light detection with 95% precision and 60 FPS.
Completed an AI Diploma (Intensive Technical Training Program) from Route Company, gaining practical exposure to Keras, TensorFlow, and PyTorch.
Built an end-to-end machine learning pipeline for credit card fraud detection, enhancing recall rates to 94%.