AI Engineer Roadmap

What the job is, how it differs from ML and data science, and where LLMs fit in AI.

1. 01What Does an AI Engineer Do?(coming soon)
2. 02AI Engineer vs ML Engineer vs Data Scientist(coming soon)
3. 03AI, Machine Learning, Deep Learning and LLMs(coming soon)

Tokens, next-token prediction, the context window, training, sampling settings and why models make things up.

1. 01What Is a Large Language Model?(coming soon)
2. 02Tokens and Tokenization(coming soon)
3. 03How an LLM Predicts the Next Token(coming soon)
4. 04Context Window: Why an AI Chat Forgets
5. 05How an LLM Is Trained: Pre-training to RLHF(coming soon)
6. 06Temperature, Top-k and Top-p(coming soon)
7. 07Why LLMs Hallucinate(coming soon)

Open vs closed, the main model families, reasoning models, and trading quality against speed and cost.

1. 01Open vs Closed Models(coming soon)
2. 02Model Families: GPT, Claude, Gemini, Llama(coming soon)
3. 03Reasoning Models vs Chat Models(coming soon)
4. 04Decision Models: When Not to Use an LLM
5. 05Choosing a Model: Quality, Speed and Cost(coming soon)

Writing prompts that work every time: roles, examples, step-by-step reasoning and output you can parse.

1. 01System, User and Assistant Messages(coming soon)
2. 02Zero-Shot vs Few-Shot Prompting(coming soon)
3. 03Chain of Thought Prompting(coming soon)
4. 04Structured Output: Getting JSON Back(coming soon)
5. 05Writing Prompts That Don't Break(coming soon)

Your first API call, streaming, what each call costs, prompt caching, and handling failures.

1. 01Your First LLM API Call(coming soon)
2. 02Streaming Responses(coming soon)
3. 03Tokens, Cost and Rate Limits(coming soon)
4. 04Prompt Caching(coming soon)
5. 05Retries, Timeouts and Fallback Models(coming soon)

Turning text into numbers that capture meaning, measuring how close two texts are, and searching millions fast.

1. 01What Are Embeddings?(coming soon)
2. 02Cosine Similarity(coming soon)
3. 03Semantic Search vs Keyword Search(coming soon)
4. 04Vector Databases(coming soon)
5. 05How Vector Indexes Work: HNSW(coming soon)

Retrieval-augmented generation: answering from your documents, and fixing it when the answer is wrong.

1. 01What Is RAG?(coming soon)
2. 02Chunking Strategies(coming soon)
3. 03Retrieval: Top-k and Metadata Filters(coming soon)
4. 04Hybrid Search: Keywords Plus Vectors(coming soon)
5. 05Reranking(coming soon)
6. 06Why RAG Fails and How to Fix It(coming soon)
7. 07RAG vs Fine-Tuning vs Long Context(coming soon)

Measuring quality instead of guessing: failure points, criteria, methods, test sets and regression checks.

1. 01Why "It Looks Good" Isn't Enough
2. 02Where LLM Apps Fail
3. 03What to Measure
4. 04Eval Methods: Code, Human and LLM-as-Judge
5. 05Building a Golden Test Set(coming soon)
6. 06Evaluating RAG: Faithfulness and Recall(coming soon)
7. 07Regression Testing Prompts(coming soon)

The loop that turns a model into an agent, how it picks tools, how it remembers, and the limits that keep it safe.

1. 01The Agent Loop
2. 02Tool Calling
3. 03ReAct: Reason, Then Act(coming soon)
4. 04Agent Memory: Short-Term and Long-Term(coming soon)
5. 05Agent Guardrails

The Model Context Protocol: one standard way to plug any tool into any agent.

1. 01What Is MCP?
2. 02Building an MCP Server(coming soon)
3. 03MCP Transports: Local vs Remote(coming soon)

When a fixed workflow beats an agent, how several agents split work, and how to judge a whole run.

1. 01Workflows vs Agents: When Not to Use an Agent(coming soon)
2. 02Multi-Agent Patterns: Orchestrator and Workers(coming soon)
3. 03How Coding Agents Work(coming soon)
4. 04Evaluating an AI Agent(coming soon)

Deciding what the model sees on every call, and the ways a crowded context makes it worse.

1. 01Prompt Engineering vs Context Engineering(coming soon)
2. 02Context Sources: What Goes Into the Window(coming soon)
3. 03Context Compaction(coming soon)
4. 04Lost in the Middle: Long-Context Limits(coming soon)
5. 05How Context Fails(coming soon)

Prompt injection, guardrails on input and output, private data, bias, and attacking your own app first.

1. 01Prompt Injection(coming soon)
2. 02Input and Output Guardrails(coming soon)
3. 03PII and Private Data in LLM Apps(coming soon)
4. 04Bias and Fairness(coming soon)
5. 05Red-Teaming an LLM App(coming soon)

Seeing what your app does in production and keeping the bill under control.

1. 01Tracing an LLM App(coming soon)
2. 02Monitoring Cost and Latency(coming soon)
3. 03Monitoring Quality in Production(coming soon)
4. 04Semantic Caching and Model Routing(coming soon)

Running models yourself: locally, at scale with batching and the KV cache, smaller with quantization.

1. 01Running an LLM on Your Laptop(coming soon)
2. 02How Inference Serving Works: The KV Cache(coming soon)
3. 03Batching and Throughput(coming soon)
4. 04Quantization(coming soon)
5. 05Self-Hosting vs API: The Cost Math(coming soon)

When to train a model on your data, how LoRA makes it cheap, and how to check it helped.

1. 01When to Fine-Tune a Model(coming soon)
2. 02LoRA: Fine-Tuning on a Budget(coming soon)
3. 03Preparing Training Data(coming soon)
4. 04Distillation: A Small Model Learns From a Big One(coming soon)

Models that see, hear and speak: images in and out, speech and real-time voice.

1. 01Vision Models: How an LLM Reads an Image(coming soon)
2. 02How Image Generation Works(coming soon)
3. 03Speech to Text and Text to Speech(coming soon)
4. 04Building a Voice Agent(coming soon)

A step-by-step method, then five classic AI systems designed from requirements to evals.

1. 01How to Approach an AI System Design Interview(coming soon)
2. 02Design a Chat-With-Your-Docs App(coming soon)
3. 03Design a Customer Support Agent(coming soon)
4. 04Design a Semantic Search Engine(coming soon)
5. 05Design an AI Code Review Bot(coming soon)
6. 06Design an LLM Gateway(coming soon)