AI Engineering

What Is AI Engineering?

• Build applications on top of existing foundation models.
• Engineering is the systematic application of established knowledge and methods to solve real-world problems, producing solutions that can be verified, operated, and maintained.

Why Did AI Engineering Emerge?

• Models become larger → general capabilities improve → pretraining costs rise → most developers cannot train frontier models themselves → providers release model APIs and open weights → application developers reuse existing models → AI engineering emerges.
  • Previously
    • Collect data → train a model → deploy the model → build an application.
  • Today
    • Define requirements → select an existing model → build an application → evaluate and improve it.

AI Application Scenarios

• Coding
• Conversational Assistants
• Information Aggregation
• Workflow Automation
• Image and Video Generation
• Writing
• Education

Technology Stack

• Application Layer
  • Prompt Engineering
  • Context Engineering
  • Retrieval-Augmented Generation (RAG)
  • AI Agents
  • Evaluation
  • User Interfaces
• Model Layer
  • Model Architecture
  • Pretraining
  • Fine-Tuning
  • Dataset Engineering
  • Inference Optimization
• Infrastructure Layer
  • GPU Computing
  • Distributed Training Infrastructure
  • Model Serving
  • Resource Scheduling
  • Monitoring and Observability