Skip to main content

Explore Problems

Certifications Quiz Pricing

⌘K

Foundations of AI Engineering

Course Contents

What Is AI Engineering? (And Why It Exploded)

The AI Stack: Application, Model & Infrastructure Layers

AI Engineer vs ML Engineer vs Full-Stack Engineer

What You Can Build: The Foundation-Model Use-Case Map

The AI Engineering Mindset (Build-vs-Buy, Eval-First, Iterate)

How This Course Works (Prerequisites & Optional Primers)

Language Models 101: Predicting the Next Token

Tokenization: How Text Becomes Numbers (BPE)

Embeddings: Meaning as Vectors

The Transformer Architecture — The Big Picture

Self-Attention, Intuitively

Multi-Head Attention & Positional Encoding

Why Transformers Won (vs RNNs/LSTMs)

Hands-On: Build a Tiny GPT (Transformers Demystified in Code)

Pretraining: Learning from the Internet

Training Data: Garbage In, Garbage Out

Model Size & Scaling Laws

The Model Families: GPT, Claude, Gemini, Llama, Qwen, Mistral, DeepSeek

Open vs Closed Models — The Tradeoffs

Base vs Instruct vs Reasoning Models

Why Raw Base Models Aren't Useful

Supervised Fine-Tuning (SFT): Teaching Instruction-Following

RLHF: Aligning with Human Preferences

DPO & Modern Preference Tuning

Reasoning Models: How o1 / R1 "Think"

From Logits to Probabilities (Softmax)

Greedy Decoding vs Sampling

Temperature, Top-k & Top-p Explained

Controlling Output: Stop Sequences, Max Tokens, Penalties

The Probabilistic Nature of AI (Determinism, Seeds)

Where Hallucinations Come From

Anatomy of an LLM API Call (Messages, Roles, Params)

Setting Up: Anthropic (Claude) & OpenAI SDKs

Running Open Models Locally with Ollama

Tokens, Context Windows & Why They Matter

Understanding API Pricing (Token Math)

Hands-On: Build a CLI Chatbot

Context Length: Long-Context Models & Their Limits

Multimodality: Text, Vision, Audio

Latency vs Cost vs Quality — The Iron Triangle

Reading Benchmarks Critically (MMLU, GPQA, SWE-bench)

A Repeatable Model-Selection Workflow

Build vs Buy & the Model Router

In-Context Learning: Zero-Shot & Few-Shot

Anatomy of a Great Prompt

System, User & Assistant Roles

Writing Clear, Explicit Instructions

Roles, Delimiters & XML Tags

Few-Shot Done Right (Choosing Examples)

Iterating on Prompts Systematically

Chain-of-Thought: Giving Models Time to Think

Self-Consistency & Majority Voting

ReAct: Reasoning + Acting (Prompt Pattern)

Prompt Chaining & Task Decomposition

Meta-Prompting & Reusable Templates

Automating Prompt Optimization (DSPy)

Prompt Versioning & Management

Why You Need Structured Output

JSON Mode & Response Schemas

Function / Tool Calling Basics

Designing Good Tool Schemas

Parsing, Validation & Retries (Pydantic)

Hands-On: Extract Structured Data from Unstructured Text

What Is Context Engineering? (The 2026 Shift)

Managing the Context Window (Selection & Compression)

Memory, State & History as Context

Prompt Injection & Jailbreaking

Defenses Against Prompt Attacks

Information Extraction & Data-Leakage Risks

Anatomy of an LLM API Call (Messages, Roles, Params)

Dissect a single LLM API call with Claude and OpenAI — the stateless-function model, system, user and assistant roles, core parameters, stop reasons and usage, and how multi-turn chat works.

Loading...

Loading content...