Master the art and science of working with large language models, from understanding how transformers work under the hood to shipping production AI applications. You will practice advanced prompting patterns like chain-of-thought, few-shot, RAG, and ReAct, explore fine-tuning with LoRA and QLoRA, and build apps with the OpenAI and Anthropic APIs, LangChain, LlamaIndex, embeddings, and vector databases. The course closes with evaluation, safety, and hallucination mitigation so what you build can be trusted.

Modules

How LLMs Work Under the Hood

Lessons

• From Next-Token Prediction to Intelligence Preview
• Attention and the Transformer Architecture
• Pretraining Instruction Tuning and RLHF

Advanced Prompting Techniques

Lessons

• Few-Shot Prompting and In-Context Learning
• Chain-of-Thought and Structured Reasoning
• ReAct and Retrieval-Augmented Generation Patterns

Fine-Tuning and Efficient Training

Lessons

• When to Fine-Tune and When to Prompt
• LoRA and Parameter-Efficient Methods
• QLoRA and Training on a Budget

Building AI Applications

Lessons

• Working with the OpenAI and Anthropic APIs
• Orchestrating Workflows with LangChain
• Document Pipelines with LlamaIndex

Embeddings and Vector Databases

Lessons

• What Embeddings Actually Represent
• Chunking Strategies for Retrieval
• Choosing and Integrating a Vector Database

Evaluation Safety and Reliability

Lessons

• Evaluating LLM Outputs Systematically
• Understanding and Reducing Hallucination
• Guardrails Red Teaming and Safe Deployment