📄 หัวข้อเทรน

ML Training & Deployment with AWS SageMaker + EC2 Docker

1. System Overview & Architecture

  • End-to-End ML Flow (Train → Version → Deploy → Inference)

  • Dev / Prod Environment Strategy

2. SageMaker Jupyter Notebook Setup

  • Project & Folder Structure

  • Environment & Dependency Management

  • Config Management ( ENV)

3. Data Ingestion & Feature Engineering

  • Data Loading (API / S3)

  • Time-series Alignment

  • Feature Selection

4. Anomaly Model Training

  • Model Selection (AutoEncoder / One-Class / Isolation Forest)

  • Training Pipeline

  • Parameter Tuning

5. Classification Model Training

  • Labeling Strategy

  • Train / Validation / Test Split

6. Model Evaluation

  • Confusion Matrix

7. Experiment Tracking & Model Versioning

  • MLflow Experiment Design

  • Parameter & Metric Logging

  • Artifact Management

  • Model Version Control (v1, v2, v3…)

8. Dockerized Inference – Anomaly Model

  • Docker Image Structure

  • Environment Variable Design

  • Inference Script Integration

  • Configurable Inference Parameters

9. Anomaly Inference Scheduling

  • Inference Interval Configuration (minutes)

  • Time Window Control (start / end time)

  • Schedule Control via Dockerfile / ENV

  • Container Restart Strategy

10. Dockerized Inference – Classification Model

  • REST API Structure

  • Health Check Endpoint

11. Deployment on EC2

  • EC2 Environment Preparation

  • Docker Build & Run

  • Container Naming & Version Tagging

12. Logging & Monitoring

  • Anomaly Event Logging

  • Container Log Inspection