Y.S. Yang

Designed and deployed production-grade generative AI applications using LLMs, RAG, vector databases, prompt engineering, and agent-based workflows to automate knowledge retrieval and business processes. Built scalable RAG pipelines combining document ingestion, semantic chunking, embeddings, hybrid search, reranking, and LLM response generation while improving retrieval accuracy and reducing hallucinations. Fine-tuned open-source language models using LoRA and QLoRA and created task-specific evaluation datasets comparing accuracy, latency, token consumption, and inference cost. Built AI agent systems capable of calling APIs, retrieving enterprise information, executing structured workflows, maintaining contextual state, and coordinating multiple specialized AI components. Developed high-performance inference APIs with FastAPI and containerized services using Docker for reliable deployment across development and production environments. Implemented automated model evaluation pipelines

Developed machine learning and generative AI solutions for enterprise applications involving natural language processing, document intelligence, classification, semantic search, and intelligent automation. Built NLP pipelines for text preprocessing, classification, entity extraction, semantic similarity, document retrieval, and structured information extraction using Python, PyTorch, and Hugging Face Transformers. Designed enterprise RAG architectures integrating embedding models, vector search, metadata filtering, reranking algorithms, and LLM inference to improve knowledge retrieval across large document collections. Evaluated multiple foundation models based on task accuracy, hallucination behavior, inference latency, context window utilization, and operational cost to identify suitable models for production workloads. Built RESTful AI microservices and integrated machine learning models using Python, FastAPI, Docker, cloud services, and asynchronous processing pipelines. Applied m

Developed machine learning models for classification, forecasting, anomaly detection, recommendation, and predictive analytics using Python, scikit-learn, TensorFlow, and PyTorch. Created reusable data preprocessing pipelines covering missing-value treatment, feature engineering, normalization, categorical encoding, data validation, and automated dataset preparation. Built deep learning projects for natural language processing and computer vision and ran systematic experiments to compare architectures, hyperparameters, training strategies, and evaluation metrics. Developed computer vision pipelines for image classification and object detection using OpenCV, convolutional neural networks, and transfer learning. Implemented NLP solutions for text classification, sentiment analysis, keyword extraction, semantic similarity, and document categorization, and built backend APIs exposing trained models to internal software applications. Containerized ML applications using Docker for cloud-bas