Generative AI
Explore how generative AI models create text, images and code, and how to apply them to real tasks.
- Fees
- Contact us for fees
About this course
This course teaches you to build useful applications on top of large language models: assistants that answer from your own documents, chatbots that call real tools, and agents that carry out multi-step tasks. You start with how these models work and where they fail, then learn the craft that makes them dependable: clear prompts, retrieval, tool use and testing.
All work is done in Python against hosted model APIs, so every module ends in something you can run and show. Accuracy, privacy, cost and safety are part of each module, not a chapter added at the end.
What you will be able to do
- Explain how large language models generate text, and recognise where they hallucinate, show bias or run out of context.
- Write and test prompts that give structured, repeatable output.
- Call model APIs from Python and handle errors, cost and rate limits.
- Build a retrieval-augmented generation (RAG) system that answers from your own documents and cites its sources.
- Build an agent that plans steps and calls tools or APIs.
- Evaluate answers with a small test set and simple scores instead of judging by eye.
- Put a working prototype behind a simple web interface and explain how it would be deployed safely.
Syllabus
10 modules. Select a module to see its topics.
Module 1: Generative AI foundations
- What generative AI is
- Text, image, code and speech models
- How a language model predicts text
- Tokens and context windows
- Strengths, limits and common business uses
Module 2: Python for AI work
- Functions, data structures and files
- JSON and HTTP APIs
- Keeping API keys out of code
- Virtual environments
- Notebooks and scripts
Module 3: Working with large language models
- Hosted and open models compared
- Chat APIs and message roles
- Temperature and other settings
- Streaming
- Cost, latency and rate limits
Module 4: Prompt engineering
- Clear instructions and examples
- Few-shot prompts
- Step-by-step reasoning prompts
- Structured JSON output and validation
- Prompt templates
- Testing and versioning prompts
Module 5: Embeddings and vector search
- What embeddings are
- Similarity search
- Splitting documents into chunks
- Vector stores such as FAISS, Chroma or pgvector
- Hybrid search basics
Module 6: Retrieval-augmented generation (RAG)
- The full pipeline
- Loading PDFs and web pages
- Improving retrieval
- Citing sources
- Common failures and fixes
Module 7: Tool calling and AI agents
- Function calling
- Planning and multi-step workflows
- Memory and state
- Agent frameworks such as LangChain or LlamaIndex, and when plain code is better
- Safety checks
Module 8: Evaluation, safety and guardrails
- Building a test set
- Scoring answers
- Spotting hallucinations
- Prompt-injection and data-leak risks
- Content filters
- Human review
- Privacy and consent
Module 9: Open models and fine-tuning
- When fine-tuning helps and when RAG is enough
- The Hugging Face model hub
- Running a small open model
- Parameter-efficient fine-tuning (LoRA) in outline
- Preparing a dataset
Module 10: Building and deploying an application
- A Streamlit or Gradio interface
- An API with FastAPI
- Logging and cost monitoring
- Deployment options
- Capstone project and demo
Tools and technologies
- Python
- Jupyter
- Hosted language-model APIs (for example OpenAI, Google Gemini or Anthropic Claude)
- Hugging Face Transformers
- LangChain or LlamaIndex
- Vector stores (FAISS, Chroma, pgvector)
- FastAPI
- Streamlit or Gradio
- Git and GitHub
Projects you will build
Document question-answering assistant: RAG over a set of PDFs, with source citations.
Support chatbot that follows a written policy and calls a function to look up an order status.
Content and summary generator, with a script that scores tone, length and accuracy.
Research agent that plans a task, searches, reads and writes a short report.
Capstone: an end-to-end assistant for a chosen business case, with a web interface, logging and a short evaluation report.
Before you start
- This outline starts from the basics: Python is taught in module 2, so no earlier programming is assumed.
- Helpful but not required: comfort with logical thinking, and a laptop with a reliable internet connection.
- Entry requirements, and any charges for AI services used in class, are confirmed by Pathway Career when you enquire.
Career paths
Examples of roles that use these skills. They are not a promise of a job.
- Generative AI developer
- AI application developer
- Prompt engineer
- LLM engineer (junior)
- Conversational AI developer
- AI solutions analyst
Questions about this course
Do I need AI or programming experience first?
This outline assumes no AI background. Python is taught from the start of the course, so comfort with logical thinking is enough to begin. Entry requirements are confirmed by Pathway Career when you enquire.
Will I build things, or only learn theory?
You build. Each module ends with working code, and the projects listed on this page are made from those pieces.
Which AI tools and models will we use?
The outline uses Python with hosted model APIs, Hugging Face models, LangChain or LlamaIndex and a vector store. Pathway Career confirms the exact providers and versions when you enquire.
Do I need to pay for AI services?
Some model APIs charge by use. Pathway Career confirms what is provided in class, and what you would set up yourself, when you enquire.
How is this different from the AI & Machine Learning course?
This course builds applications on top of ready-made models. AI & Machine Learning teaches how to train models from data.
Is it only for software developers?
No. Analysts, testers and content or support teams use generative AI too, but you will write Python code in this course.
Next step
Want the syllabus, schedule and fees for Generative AI? Send an enquiry and ask for the full course outline.
Send an enquiry