Generative AI for Software Developers
Generative AI for Software Developers is a course in using English prompts with ChatGPT and Claude across the software lifecycle, from system design, coding, legacy code, debugging and testing to working with GitHub Copilot, coding agents, LLM APIs, RAG and MCP. It suits developers, tech leads and freelancers who want to use AI safely and effectively.
- Duration18 hours · 3 days
- FormatOnsite / live online
- Next round14–16 Sep 27
- CertificateIncluded
Course overview
Generative AI has become part of everyday software development, from system design, coding and bug fixing to writing tests and documentation. But the results depend on how clearly developers communicate with the AI, whether they give it enough context and how they check its work. Developers who use AI without a method often end up with code that looks right but hides mistakes, or accidentally share sensitive data.
This course shows developers how to use generative AI systematically across the software development lifecycle. It starts with how LLMs work, writing English prompts and managing context with ChatGPT and Claude, then uses AI to design systems, databases and APIs, handle legacy code, debug and write tests, work with GitHub Copilot in the IDE and collaborate with coding agents. Learners then build UI prototypes, call LLM APIs from code, create a RAG chatbot and connect tools through MCP, with security and verification at every step. (3 days, 6 hours per day, 18 hours in total, Intermediate level.)
What you’ll gain
- Understand how LLMs work and choose the right AI tool for each kind of development task
- Write English prompts and manage context so the AI produces what you need
- Use AI to help design systems, databases, APIs and diagrams
- Use AI to explain, refactor and handle legacy code safely
- Use AI to help debug and write unit tests
- Work with GitHub Copilot and coding agents and review their output
- Call LLM APIs from code and build a RAG chatbot and a simple MCP server
- Follow good practice on security, copyright and personal data when using AI
Who this course is for
- Software developers working on the frontend, backend or full stack
- Tech leads and software architects who set the team's approach to AI
- Freelance and startup developers who want to build prototypes faster
- QA and DevOps engineers who want AI to help with testing and automation
- Developers maintaining older systems who want AI to help read and improve code
Prerequisites
- Programming experience in at least one language, such as Python, JavaScript, TypeScript or Java
- Understanding of basic software concepts such as APIs, databases and Git
- Basic reading ability in English; fluency is not required
- A laptop that can run VS Code, Git and Python, plus your own ChatGPT, Claude and GitHub accounts
Curriculum
Course Details
This is the 2026 edition that replaces the previous content of Generative AI for Software Developer. It keeps the original core of using English prompts with ChatGPT and Claude for system design, writing and analysing code, handling legacy code and building prototype websites and chatbots, and adds context engineering, debugging and unit testing, GitHub Copilot agent features, coding agents such as Claude Code and Codex, calling LLM APIs, RAG and
MCP. It runs for 3 days, 6 hours per day (18 hours in total, 09:00-16:00), as lectures with a lab in every topic. Intermediate level. Learners use their own accounts and API keys; some tools need a paid plan or API credits, paid for by the learner. Teams that want to roll out coding agents across the whole lifecycle can continue with AI Coding Agent for SDLC.
Day 1 Generative AI Foundations and Prompting for Developers
Section 1: Generative AI for Developers in 2026
- How LLMs work: tokens, context windows, transformers and reasoning models
- How chat assistants, coding assistants and coding agents differ
- The landscape: ChatGPT, Claude, Gemini, GitHub Copilot, Claude Code and Codex
- Strengths, limitations and hallucinations every developer should know
- Lab: compare several models on the same coding task
Section 2: Lab: Setting Up Tools and Accounts Safely
- Set up ChatGPT and Claude with your own accounts and use Projects to keep context
- Install GitHub Copilot in VS Code
- Create API keys and store them safely in environment variables
- Privacy settings, data use and company policy before sharing code with AI
- Costs, quotas and tracking API usage
Section 3: Lab: English Prompts for Developers
- Prompt structure: role, task, context, constraints and output format
- English vocabulary and sentence patterns that give AI clear instructions
- Few-shot examples and asking the AI to ask clarifying questions
- Iterating on prompts and asking the AI to check its own answer
- Lab: write prompt templates to generate, refactor and explain code
Section 4: Context Engineering
- Give the AI code, files and errors in the right amount
- Project instruction files: AGENTS.md, CLAUDE.md and copilot-instructions.md
- Ask for structured output such as JSON, tables and diffs
- Manage long conversations and reset context when answers drift
- Lab: write instruction files for a sample project
Section 5: Lab: Writing and Explaining Code with AI
- Generate functions and classes from requirements
- Have the AI explain code line by line and translate it between languages
- Create SQL, regular expressions and automation scripts from prompts
- Check the quality of AI-generated code before using it
- Lab: build a sample module with explanations and documentation
Day 2 AI for Design, Code and Coding Agents
Section 6: Lab: Designing Systems and Databases with AI
- Turn requirements into user stories and acceptance criteria
- Have the AI propose architecture options with trade-offs
- Design an ERD and database schema
- Design a REST API with OpenAPI and draw diagrams with Mermaid
- Lab: design a meeting room booking system from requirements to API
Section 7: Lab: Handling Legacy Code with AI
- Have the AI explain undocumented legacy code
- Add comments, docstrings and a README
- Refactor step by step safely, guarded by tests
- Reverse-engineer code into diagrams and plan a version migration
- Lab: read, summarise and improve undocumented Java or Python code
Section 8: Lab: Debugging and Unit Testing with AI
- Give the AI stack traces and logs to find the root cause
- Ask the AI for hypotheses and ways to test them instead of guessing fixes
- Generate unit tests and edge cases
- Use AI with test-driven development
- Lab: fix bugs in a sample app and add tests to prevent regressions
Section 9: Lab: AI Coding Assistants in the IDE
- GitHub Copilot code completion and next edit suggestions
- Copilot Chat for questions about the workspace
- Agent mode for multi-file edits and commands, reviewing diffs before accepting
- Choose a model for the task and use custom instructions
- Lab: add a feature to a sample app with Copilot
Section 10: Coding Agents: Claude Code and Codex
- How terminal-based coding agents work
- Plan before acting, and set permissions for editing files and running commands
- Work with Git: branches, commits and pull requests
- Assign an issue to an agent on GitHub and review the pull request
- Lab: have an agent build a small feature and review the result
Day 3 Building Prototypes and AI Apps Safely
Section 11: Lab: Prototyping and UI with AI
- Create web pages with HTML and Tailwind CSS from prompts
- Build UI from a sketch or screenshot
- Preview in chat with Claude Artifacts or ChatGPT Canvas
- Deploy a static site to GitHub Pages or Vercel
- Lab: build a prototype landing page and put it online
Section 12: Lab: Calling LLM APIs from Code
- Call the OpenAI Responses API and Anthropic Messages API with the Python SDKs
- System prompts, temperature and managing conversation history
- Structured output with JSON Schema and Pydantic
- Streaming, error handling and cost control
- Lab: build a pull request summariser with an LLM API
Section 13: Lab: Chatbots and RAG
- RAG concepts: chunking, embeddings and vector search
- Build a knowledge base from project documents
- Cite sources in answers and say 'I don't know' when data is missing
- Test the quality of chatbot answers
- Lab: an FAQ bot for development team documents
Section 14: Tool Calling and the Model Context Protocol (MCP)
- Tool calling: letting the model call your functions
- What MCP is: tools, resources and prompts
- Connect MCP servers to Claude Code or VS Code
- The risks of third-party tools and MCP servers
- Lab: build a simple MCP server that lets AI look up project data
Section 15: Workshop: Best Practices and Capstone
- Security: secrets, prompt injection and AI-suggested dependencies
- Copyright, code licences and PDPA when using AI with real data
- Fit AI into the dev lifecycle: plan, build, test, review and deploy
- Capstone: build a prototype app with a chatbot from design to deployment
- Present the work and collect proven prompts into a personal prompt library
Schedule & training options
For individuals — public rounds
For organisations — in-house / private
- Tailor the content to your team’s tools and projects
- Your dates, at your office or live online
- Quotation with tax ID for procurement
Instructors
Frequently asked questions
How do I write prompts that get good code from AI?
State the role, the task, the context, the constraints and the output format, for example the language, library versions, related code and the tests it must pass. Give examples when a specific pattern is needed and ask the AI to ask clarifying questions when the task is unclear. For larger work, have the AI plan first, work in small steps and check the result by running tests every time. The course provides English prompt templates for writing, refactoring and explaining code.
How can AI help read and refactor legacy code?
Start by having the AI explain the code piece by piece and summarise it as diagrams to see the big picture. Then have it write tests that pin down the current behaviour, and refactor in small steps, running the tests each time. Avoid asking the AI to rewrite the whole system at once, and make sure the code you share contains no secrets or data your organisation does not allow to leave.
What is RAG (retrieval-augmented generation)?
RAG retrieves relevant information from your own documents first and passes it to the LLM as grounding for its answer. Documents are split into chunks, turned into embeddings, stored in a vector database and searched by meaning. The AI can then answer from your organisation's current information, cite its sources and make things up less often. On the course you build an FAQ bot from development team documents.
Who is the Generative AI for Software Developers course for, and what should I know beforehand?
It suits frontend, backend and full-stack developers, tech leads, software architects, QA and DevOps engineers, and freelance developers. You should program in at least one language, understand APIs, databases and Git, read basic English, and bring your own ChatGPT, Claude and GitHub accounts.
What will I be able to do after the Generative AI for Software Developers course?
You will be able to write prompts and manage context effectively, use AI to design systems, databases and APIs, handle legacy code, fix bugs and write tests, work with GitHub Copilot and coding agents while reviewing their output, call LLM APIs from code, build a RAG chatbot and a simple MCP server, and keep your own prompt library. Participants who complete the training receive a Certificate of Completion.
Can the Generative AI for Software Developers course be run in-house for our team?
Yes. IT Genius Institute runs in-house training and can adapt the lab examples to the languages, frameworks and AI tools your team uses, and help shape AI usage guidelines that fit your organisation's security and data policies. Contact us for a quotation and to discuss the content.