Course Overview
You will write custom GPTs for AI-assisted application development. You will gain a clear understanding of prompting techniques, building/configuring a custom GPT, and end-to-end application development with AI assistance. With over 39 labs and lectures, this hands-on intensive course is ideal for anyone needing to develop applications with AI assistance. Direct access to the AI platform is not required, as all traffic is managed through the training provider. An OpenAI Plus subscription is required, as the course content utilizes ChatGPT Premium features.
Who should attend
- Application Developers
- Project Managers
- System Engineers
- DevOps Management
Prerequisites
- Previous exposure to any programming language, preferably Python
- Experience writing prompts, or previous prompt engineering training/experience helpful, but not required
Course Objectives
- Develop and Write Code with AI
- Plan and Scope a Project with AI
- Build and Configure a Custom GPT
- Fine-Tune a Custom GPT
- Deploy an Application with AI Assistance
- Understand Large Language Models and Prompting
- Deploy Advanced Prompting Techniques
- Learn Complex Programming Concepts with AI
- Develop Instructions for a Custom GPT
- Defining Prompts and Parameters
- Introduction to AI-Assisted Application Development
Outline: Accelerating Software Development with GitHub Copilot (ADAI)
Getting Started (Assumed Proficiency, Fast Ramp)
- Course Scope and Expectations
- GitHub Copilot Account Verification
- Editor Readiness Check (Vim, VS Code, JetBrains)
Mental Model: How Copilot Actually Works
- GitHub Copilot Is Not Intelligent
- What Copilot Reads: Files, Tabs, and Context
- Why Copilot Behaves Like Autocomplete Without Discipline
Context Staging: Teaching Copilot What Matters
- Context Staging Principles
- Lab: Staging Interfaces, Schemas, and Examples
- Lab: Reducing Noise to Improve Output Quality
Comment-First Specification (Speed)
- Comment-First Development Discipline
- Writing Inputs, Outputs, Rules, and Failure Modes
- Lab: Generating Code from Full Comment Specifications
Iterative Refinement Without Thrash
- The Ghost in the Machine Technique
- Why Regeneration Is a Smell
- Lab: Incremental Refinement Inside the Same File
Accuracy Through Constraints
- Preventing Hallucinations with Constraints and Invariants
- Using “Must”, “Must Not”, and Non-Negotiables
- Lab: Fixing Incorrect Output Using Constraints Only
Working in Existing Codebases
- Why Greenfield Demos Lie
- Inline Completions vs Copilot Chat
- Lab: Explaining Unfamiliar Code with Copilot Chat
- Lab: Making Targeted Changes with Inline Completions
Testing and Validation with Copilot
- Developer Responsibility and Verification
- Using Copilot to Generate Tests (Safely)
- Lab: Generate Tests to Validate a Copilot-Assisted Change
Cross-Language and Cross-Platform Discipline
- Constraint Patterns Across Languages
- Procedural vs Object-Oriented Guidance
- The Rosetta Stone Technique
- Lab: Translating Intent Across Platforms Without Migration
Staying Fast Without Creating Risk
- What Copilot Logs and Does Not Log
- Safe vs Unsafe Usage Patterns
- What Must Never Be Delegated to Copilot
- Team-Level Rules for Sustainable Speed