A practical MLOps project deploying a fine-tuned DistilBERT sentiment model with a FastAPI service and optional dynamic quantization, cutting latency by 60%. Includes CI/CD (GitHub Actions), automated testing, and performance logging — showcasing production-ready ML deployment.