Pebellslearn by building
RoadmapsMatrixCoursesSandboxSign in

Course details

AI Engineer

Learn how language models work and how to build dependable products on them. You will reason about tokens and context, write prompts that hold up, connect models to data and tools, and keep the whole thing safe — all while building one bike shop's support assistant.

Ready to learn5 sections · 22 lessons
Start learning

Course curriculum

Introduction to AI Engineering5 lessons

Meet the AI engineer's role, see what a large language model actually does, separate training from inference, and learn the two ideas behind most real products: embeddings and retrieval-augmented generation.

  1. What Is an AI Engineer?12 min
  2. Large Language Models12 min
  3. Training, Fine-tuning and Inference12 min
  4. Embeddings and Vector Databases12 min
  5. Retrieval-Augmented Generation (RAG)15 min
Tokens, Context and Sampling4 lessons

Count in tokens, budget a context window, and set the sampling dials — temperature, top-k, top-p and repetition penalties — so the assistant is precise where it must be and varied where it may be.

  1. Tokens — How Models Read and Bill12 min
  2. The Context Window12 min
  3. Temperature, Top-k and Top-p12 min
  4. Repetition Penalties12 min
Working with Model APIs4 lessons

Shape behaviour with system prompts, get output your code can trust with structured formats, let the model act through function calling, and make responses feel fast and cost less with streaming and prompt caching.

  1. System Prompts and Message Roles12 min
  2. Structured Output12 min
  3. Function Calling and Tools12 min
  4. Streaming and Prompt Caching15 min
Prompting Techniques and Context Engineering4 lessons

Teach by example with few-shot prompts, get better multi-step answers with chain of thought, decide what goes into the model's context, and prove each change helps with a small evaluation set.

  1. Zero-Shot and Few-Shot Prompting12 min
  2. Chain-of-Thought Prompting12 min
  3. Context Engineering12 min
  4. Evaluating Prompts and Context15 min
AI Safety and Ethics5 lessons

Defend against prompt injection, protect secrets and personal data, test for unfair treatment, guard what goes in and comes out, and attack your own app before anyone else does.

  1. Prompt Injection12 min
  2. Security and Privacy12 min
  3. Bias and Fairness12 min
  4. Guarding Inputs and Outputs12 min
  5. Adversarial Testing15 min