amitshekhariitbhu/ai-engineer-roadmap

AI Engineer Roadmap

AI Engineer Roadmap - A step-by-step AI Engineering roadmap to become an AI Engineer, with a blog for every topic

> This roadmap is helpful for roles such as:

> - AI Engineer

- Gen AI Engineer

- LLM Engineer

- Agentic AI Engineer

- AI Agent Engineer

- Forward Deployed Engineer

- AI Solutions Architect

- AI Platform Engineer

- Applied AI Engineer

- MLOps Engineer

- LLMOps Engineer

Table of Contents

• How to Use This Roadmap
• Step 0: Must Know
• Step 1: Machine Learning Foundations
• Step 2: Deep Learning and Neural Networks
• Step 3: Generative AI and the Transformer Architecture
• Step 4: How LLMs Generate Text
• Step 5: Modern LLM Architecture
• Step 6: Types of Language Models
• Step 7: Training, Fine-Tuning, and Alignment
• Step 8: Prompt Engineering and Context Engineering
• Step 9: Vector Search and Retrieval-Augmented Generation (RAG)
• Step 10: AI Agents and Agentic Systems
• Step 11: Agentic Engineering and Agent Frameworks
• Step 12: LLM Inference Engineering
• Step 13: Evaluation and Observability
• Step 14: AI Safety and Security
• Step 15: Multimodal AI and Generative Models
• Step 16: AI Infrastructure, Deployment, and System Design
• Step 17: Frontier Ideas in AI
• Step 18: Prepare for AI Engineering Interviews

Prepared and maintained by the Founder of Outcome School: Amit Shekhar

Follow Amit Shekhar

• X/Twitter
• LinkedIn
• GitHub

Follow Outcome School

• YouTube
• X/Twitter
• LinkedIn
• GitHub

I teach at Outcome School

• AI and Machine Learning

---

Note: We will keep updating this roadmap with new topics and blogs.

---

How to Use This Roadmap

This roadmap is designed so that anyone can follow it in the given order, even without any prior background in AI.

• Follow the steps in order. Each step builds on top of the previous one.
• Inside each step, read the blogs in the given order. Every blog explains one concept in simple words with examples.
• Do not skip Step 1 and Step 2, even if you are in a hurry. Everything else in AI Engineering is built on top of them.
• If you already know a topic, move to the next one.
• Every blog is free to read and is written for beginners. No jargon. No assumptions.

Let's get started.

Step 0: Must Know

Before jumping into the details, we must know the six words that come up in every AI Engineering conversation:

• LLM
• RAG
• MCP
• Agent
• Fine-tuning
• Quantization

Learn about all six in one video: AI Engineering Explained: LLM, RAG, MCP, Agent, Fine-Tuning, Quantization

Now, we know the big picture. In the next steps, we will learn each of these in depth, one concept at a time.

Step 1: Machine Learning Foundations

In this step, we will learn what Machine Learning is, the different ways a machine can learn, and the basic terms we will keep using in every later step.

• What is Machine Learning?
• Supervised vs Unsupervised Learning
• Linear Regression vs Logistic Regression
• What is Feature Engineering in Machine Learning?
• Precision vs Recall
• What are L1 and L2 Loss Functions?
• What is Regularization in Machine Learning? L1 vs L2 Explained
• What is Reinforcement Learning?
• What is Contrastive Learning?

Step 2: Deep Learning and Neural Networks

In this step, we will learn how a neural network actually learns. We will understand the math behind gradient descent and backpropagation step by step, and the techniques that make training stable.

• What is Bias in Artificial Neural Network?
• How Does Gradient Descent Work?
• How Does Backpropagation Work? The Math Explained Step by Step
• What is Cross-Entropy Loss?
• What is Dropout in Neural Networks?
• Batch Normalization vs Layer Normalization
• What is RMSNorm? Root Mean Square Layer Normalization Explained
• What is a Recurrent Neural Network (RNN)?
• How does PyTorch work?
• How does TensorFlow work?

Step 3: Generative AI and the Transformer Architecture

In this step, we will learn what Generative AI is and how the Transformer, the architecture behind every modern LLM, works from the inside. We will go from tokens to embeddings to attention, one piece at a time.

• What is Generative AI?
• Inside ChatGPT: What Happens After You Hit Enter
• What are Autoregressive Models?
• What is Byte Pair Encoding (BPE) in LLMs?
• Tokenization in Large Language Models (LLMs) (Video)
• What are Embeddings?
• Embeddings in Machine Learning (Video)
• How do RNNs and Transformers differ?
• How Does the Transformer Architecture Work?
• Encoder vs Decoder in Transformers
• What is Self Attention in Transformers?
• How Does Attention Work? The Math Behind Q, K, and V
• Why Do We Scale Attention by √dₖ?
• What is Causal Masking in Attention?
• What is Multi-Head Attention in Transformers?
• What is Cross Attention in Transformers?
• Positional Embeddings in LLMs
• What is RoPE (Rotary Position Embedding)? The Math Behind It
• What is the Feed-Forward Network in LLMs?

Step 4: How LLMs Generate Text

In this step, we will learn how an LLM picks the next token, how we control its creativity, how the output reaches the user token by token, and where the context window fails.

• How does Temperature control LLM output?
• How do Top-k and Top-p Sampling work?
• How does Token Streaming work?
• Why is the context window limited in LLMs? (Video)
• What is the Lost in the Middle Problem in LLMs?

Step 5: Modern LLM Architecture

In this step, we will learn the improvements that modern LLMs add on top of the basic Transformer to become bigger, faster, and able to handle longer inputs. At the end, we will see all of these ideas together inside a real model.

• What is Mixture of Experts (MoE)?
• What is Grouped Query Attention (GQA)?
• How does Sliding Window Attention work?
• How do Attention Sinks work?
• What is Flash Attention and Why Is It So Fast?
• What is DeepSeek-V4 and How Does It Work? Architecture Explained

Step 6: Types of Language Models

In this step, we will learn that not every language model is a large, text-generating LLM. We will see the smaller, reasoning, recursive, diffusion-based, and decision-only models and when to use which one.

• What are Small Language Models (SLMs)?
• What are Large Reasoning Models (LRMs)?
• What are Recursive Language Models (RLMs)?
• How do Diffusion Language Models (DLMs) work?
• Jev and System One Models Explained

Step 7: Training, Fine-Tuning, and Alignment

In this step, we will learn how a pre-trained model is adapted to our own task, how it is made smaller, and how it is taught to follow instructions and human preferences.

• How does Fine-Tuning work?
• What is LoRA (Low-Rank Adaptation)?
• How does Prefix Tuning work?
• How does Knowledge Distillation work?
• What is Continual Learning in LLMs? Solving Catastrophic Forgetting
• What is Deep RL from Human Preferences? The Paper That Started RLHF
• What is InstructGPT? How GPT-3 Learned to Follow Instructions
• What is RLHF? Reinforcement Learning from Human Feedback Explained
• What is Proximal Policy Optimization (PPO)?
• What is Direct Preference Optimization (DPO)?
• What is Group Relative Policy Optimization (GRPO)?

Step 8: Prompt Engineering and Context Engineering

In this step, we will learn how to talk to an LLM so that it gives better answers, and how to manage everything that goes into its context window.

• How does Chain-of-Thought (CoT) Prompting work?
• How does Prompt Chaining work?
• How does Prompt Caching work?
• What is Context Engineering?
• How does Context Compaction work?

Step 9: Vector Search and Retrieval-Augmented Generation (RAG)

In this step, we will learn how to give an LLM knowledge that it was never trained on. We will start with how vectors are stored and searched, then move to retrieval techniques, and finally to the advanced forms of RAG.

If you want a quick introduction to RAG before starting, watch: AI Engineering Explained: LLM, RAG, MCP, Agent, Fine-Tuning, Quantization

• How does a Vector Database work?
• How does Approximate Nearest Neighbor (ANN) Search work?
• How does Semantic Search work?
• How does Hybrid Search work?
• How does a Reranker work?
• What is ColBERT? Late Interaction Retrieval Explained
• How to Chunk Documents for RAG? Chunking Strategies Explained
• How does HyDE work in RAG?
• How does an Embedding Cache work?
• How does Semantic Caching work?
• What is Agentic RAG?
• What is GraphRAG? How Knowledge Graphs Improve RAG
• What is Vectorless RAG? RAG Without Embeddings or a Vector Database

Step 10: AI Agents and Agentic Systems

In this step, we will learn how an LLM goes from answering questions to actually doing work. We will start with a single agent, see how it uses tools and memory, and then move to systems where many agents work together.

• What is an AI Agent? How It Works
• How does Function Calling work in LLMs?
• What is an AI Agent Loop?
• What is a ReAct Agent?
• What is a Plan-and-Execute Agent?
• What is a Reflection Agent?
• How does AI Agent Memory work?
• What is MCP (Model Context Protocol)?
• What are Agent Skills?
• What is OKF (Open Knowledge Format)?
• What are Multi-Agent Systems?
• What are AI SubAgents?
• How AI Agents Communicate
• What is AI Orchestration?
• What is Sakana Fugu? The Technical Report Explained
• How do Computer-Use Agents work?

Step 11: Agentic Engineering and Agent Frameworks

In this step, we will learn the engineering practices for building reliable agents, and then see how the popular frameworks and coding agents are built.

• What is Harness Engineering?
• What is Loop Engineering?
• What is Graph Engineering?
• AI Is Only as Good as Our Definition of Done
• How does LangChain work?
• How does LangGraph work?
• How does Claude Code work?
• How does Cursor work?

Step 12: LLM Inference Engineering

In this step, we will learn how to make LLMs faster and cheaper to run. We will start with what happens during inference, then learn the caching, batching, and speculation techniques, then quantization, and finally the serving engines that put it all together.

• LLM Inference Optimization
• LLM Inference Optimization (Video)
• Prefill vs Decode: LLM Inference Optimization
• The First-Token Latency Problem in LLMs (Video)
• What is Prefill-Decode Disaggregation in LLM Inference?
• What is KV Cache in LLMs?
• What is KV Cache Compression?
• What is Paged Attention in LLMs?
• What is Continuous Batching in LLMs?
• What is Speculative Decoding?
• What is N-gram Speculation in LLMs?
• What is Medusa? Multi-Head Speculative Decoding Explained
• What is EAGLE? Feature-Level Speculative Decoding Explained
• How does Model Quantization work?
• How does GGUF work?
• How does llama.cpp run LLMs on everyday hardware?
• How does vLLM work?
• How does SGLang work?
• How does TensorRT-LLM work?

Step 13: Evaluation and Observability

In this step, we will learn how to measure whether our LLM and our agent are actually doing a good job, and how to see what they are doing in production.

• What is LLM Evaluation? Metrics, Benchmarks, and Methods Explained
• What is LLM as a Judge?
• How to Evaluate AI Agents? Metrics, Methods, and Best Practices
• What is AI Agent Observability? Traces, Spans, and Metrics Explained

Step 14: AI Safety and Security

In this step, we will learn how to keep an LLM application safe, how attackers try to break it, and how AI-generated text can be identified.

• How do LLM Guardrails work?
• What is Prompt Injection in LLMs and How Do We Defend Against It?
• How does LLM Watermarking work?

Step 15: Multimodal AI and Generative Models

In this step, we will learn how AI works with images and other types of data, and the generative models that create images from noise.

• What is Multimodal AI?
• What is a Vision Transformer (ViT)?
• How do Image Embeddings work?
• What are Diffusion Models and How Do They Generate Images?
• What are Generative Adversarial Networks (GANs)?
• What are Variational Autoencoders (VAEs)?

Step 16: AI Infrastructure, Deployment, and System Design

In this step, we will learn the hardware that runs AI models, where to deploy a model, how to send each request to the right model, and how to design a complete AI system end to end.

• How does a GPU work for Deep Learning?
• How does a Google TPU work?
• How does an LPU work?
• Cloud vs On-device Model Deployment
• Android TensorFlow Lite Machine Learning Example
• What is LLM Routing? How to Send Each Query to the Right LLM
• Design a Real-Time Voice AI Agent

Supporting reads for AI System Design:

• What is System Design?
• HTTP Request vs HTTP Long-Polling vs WebSocket vs Server-Sent Events
• How do Voice and Video Call work?

Step 17: Frontier Ideas in AI

In this step, we will learn the ideas that are shaping the future of AI, from models that learn an internal picture of the world to systems that improve themselves.

• What is Joint Embedding Predictive Architecture (JEPA)?
• How do World Models work?
• What is Recursive Self-Improvement (RSI)?

Step 18: Prepare for AI Engineering Interviews

Now that we have learned everything from Machine Learning foundations to AI Ag