Course · Generative AI
Everything behind the text box, made concrete.
An 8-week generative AI course covering LLMs, embeddings, RAG, fine-tuning, multimodal models, and safety — with working code at every stage.
- 8 weeks
- Beginner friendly
- RAG
- Fine-tuning
- Multimodal
- Duration
- 8 weeksDuration
- Effort
- 6–8 hours per weekEffort
- Level
- Beginner to IntermediateLevel
- Modules
- 8Modules
The premise
Generative AI is taught either as vocabulary or as API calls, and neither survives contact with a real requirement. This course sits in between: enough of the mechanism to predict how a system will fail, and enough hands-on building to ship one that does not.
Who it is for
- Developers new to LLMs
- Data scientists moving from classical ML
- Students taking a generative AI elective
- Product and technical leads who need working knowledge
Prerequisites
- Python fundamentals — functions, files, packages
- Willingness to read documentation
- No deep learning background required
Outcomes
What you can do at the end.
Explain what a transformer does without arm-waving, and why context length costs money
Write prompts that produce structured, parseable output reliably
Build a retrieval pipeline and diagnose which stage is losing the answer
Decide between prompting, retrieval, and fine-tuning on evidence
Fine-tune a small open model with LoRA on a single GPU
Apply safety, privacy, and disclosure practices to a shipped feature
Syllabus
8 modules, in this order.
How generative models work
6 hours
- Tokens, attention, and the transformer in plain terms
- Sampling: temperature, top-p, and determinism
- Context windows, and what long context really costs
Working with LLM APIs
6 hours
- Provider APIs, streaming, and retries
- Structured output and schema validation
- Model selection by capability, latency, and price
Prompt engineering
6 hours
- Few-shot, decomposition, and self-consistency
- System prompts and role design
- Prompt injection and why it is a security problem
Embeddings and vector search
6 hours
- What an embedding encodes, and what it drops
- Similarity metrics and index types
- Chunking strategies that survive real documents
Retrieval-augmented generation
8 hours
- The full pipeline, then breaking each stage on purpose
- Hybrid retrieval and reranking
- Citations, grounding, and refusal behaviour
Fine-tuning and adaptation
8 hours
- When fine-tuning beats prompting and retrieval
- Dataset construction and quality control
- LoRA and QLoRA on a single GPU
- Quantisation for cheaper inference
Multimodal
6 hours
- Vision-language models and document understanding
- Speech in and speech out
- Image generation and its production constraints
Evaluation, safety, and cost
6 hours
- Building an eval set before you build the feature
- Hallucination measurement and mitigation
- Privacy, PII handling, and disclosure
Capstone
Build a grounded assistant over a document corpus of your choice: ingestion, retrieval, cited answers, an evaluation set with measured accuracy, and a written note on its failure modes.
Tools and stack
- Python
- Hugging Face Transformers
- Sentence Transformers
- FAISS / Chroma
- LangChain
- Ollama
- PEFT / LoRA
- Gradio
Delivery
How it can run.
Live cohort
Self-paced
Corporate batch
Campus semester
Corporate batches run against your own stack and swap the capstone for an internal use case. Campus delivery includes syllabus mapping, rubrics, and question banks — and faculty training if you want to own it afterwards.
FAQ
Generative AI — questions
Who is this generative AI course for?
Developers and technical people who can write Python and now need to build with LLMs. It assumes no deep learning background but does assume you will run the code rather than watch it.
Do I need a GPU?
Not for most of the course — API-based and CPU-friendly labs cover six of the eight modules. The fine-tuning module uses a free Colab T4, and a local Ollama path is provided for anyone working offline.
Is this course only about ChatGPT-style text models?
No. Text is the spine, but the course covers embeddings, vision-language models, speech, and image generation, because production systems mix them and the trade-offs differ per modality.
How is this different from the Agentic AI course?
This course covers the models and the retrieval layer — the raw capability. The Agentic AI course is about systems that act: loops, tools, memory, and coordination. Most people take generative AI first.
Can this run as a semester course at a university?
Yes. Forty-eight contact hours maps cleanly onto a three or four credit elective, and we supply assessment rubrics, question banks, and lab sheets alongside the syllabus mapping.
Will the material be outdated in six months?
Model names will be. The mechanisms — tokenisation, retrieval, evaluation, cost structure — have been stable for years and are what the course is built around. Cohort material is refreshed each run.
Other courses
12 weeks
Full Stack AI Engineering
A 12-week full stack AI course: build, deploy, and operate an AI application end to end — API, retrieval, agents, frontend, infrastructure, and monitoring.
8 weeks
Agentic AI
An 8-week agentic AI course on tool use, reasoning loops, memory, multi-agent coordination, MCP, and the evaluation that keeps autonomous systems trustworthy.
6 weeks
Forward Deployed AI Engineer
A 6-week course for forward deployed AI engineers: scoping in the field, prototyping against messy customer data, integration, demos that convince, and handover that lasts.
Run this
for your people.
Tell us the cohort size, the hours available, and whether this is a corporate batch or a campus program. Scope and price come back in writing within two working days.
Request a proposal