0% complete0 of 63 tasks done
🌱

Foundation Track

Start with LLM basics, prompting, and AI-assisted engineering workflows.

Phases 1-2 · Beginner
🏗️

Architect Track

Design agentic workflows and finish with portfolio-ready projects.

Phases 7-8 · Advanced

Interactive roadmap

Choose a phase and start building

01

4-5 days · Beginner

💡 Fundamentals

Build a solid foundation before touching any AI tooling. Everything advanced rests on what you learn here.

0/13tasks complete

What you will achieve

  • Explain in plain language how an LLM turns a prompt into a response.
  • Write structured prompts using at least five distinct techniques.
  • Choose an appropriate model for a given task, cost, and privacy constraint.

Understanding LLMs

4 tasks

Learn what Large Language Models are and how they work. Grasping these concepts lets you write better prompts and control costs.

TokensContext WindowLLM InferenceReasoning Models vs LLMs

Prompt Engineering

5 tasks

Prompts are your primary interface with AI. Master these techniques for dramatically better results.

Role-based promptingContext-based promptingFew-shot promptingShort promptingChain of thought

LLM Landscape

4 tasks

Know the types of models available so you can pick the right one for any task.

Frontier modelsOpen-weight modelsCoding modelsLocal models
Resources & GitHub links
CourseNeural Networks: Zero to Hero (Andrej Karpathy)github.comBuilds a GPT from scratch in Python and gives a strong mental model of how neural networks and LLMs work. Video3Blue1Brown — Neural Networks & Transformers3blue1brown.comVisual explanations for attention, embeddings, and neural network intuition. ToolOpenAI Tokenizerplatform.openai.comPaste text and see how it is split into tokens to understand context and cost. PaperAttention Is All You Needarxiv.orgThe original Transformer paper behind modern LLMs. GuidePrompt Engineering Guidepromptingguide.aiOpen-source reference for prompt techniques and examples. CourseChatGPT Prompt Engineering for Developersdeeplearning.aiShort course by DeepLearning.AI and OpenAI focused on developer prompting workflows. DocsAnthropic — Prompt engineering overviewdocs.claude.comProvider documentation with practical prompt design guidance. DocsOpenAI — Prompt engineering best practicesplatform.openai.comOfficial strategies and tactics for improving model outputs. CourseGenerative AI for Beginners (Microsoft)github.comOpen-source course surveying GenAI concepts, patterns, and code examples. ToolHugging Face Models Hubhuggingface.coBrowse open models and read model cards. ToolLMArena model leaderboardlmarena.aiCompare models using human preference and arena-style rankings.
Certifications to consider
CertNVIDIA-Certified Associate: Generative AI & LLMsnvidia.com CertMicrosoft Certified: Azure AI Fundamentals AI-900learn.microsoft.com CertAWS Certified AI Practitioneraws.amazon.com
Milestone

You can describe tokens, context windows, and inference to a colleague, and reliably improve an output by rewriting its prompt.

02

2-3 days · Beginner

⚡ AI-Powered Engineering

Start using AI tools to enhance your daily engineering workflow before you write a single line of AI code yourself.

0/6tasks complete

What you will achieve

  • Use an AI coding assistant fluidly inside a real project.
  • Know the practical limits of today's coding agents.
  • Build one automated workflow that includes an AI step.

Chat & Coding Agents

3 tasks

Get hands-on with the everyday tools before going deeper.

ChatGPTGoogle GeminiGitHub CopilotClaude CodeVS Code AI

Workflow Automation

3 tasks

Learn how AI fits into automated, multi-step pipelines.

DAGsn8nsim.aiWebhooks
Milestone

You have shipped at least one real change with an AI assistant and built one small automated workflow end-to-end.

03

3-4 days · Intermediate

🔐 AI APIs & Security

Move from chat interfaces to programmatic access. Learn to call AI models like any other API safely.

0/8tasks complete

What you will achieve

  • Call an LLM API from the terminal and from code.
  • Keep API keys and rate limits secure by default.
  • Route requests across models to balance cost and capability.

API Interaction

3 tasks

The same request, two ways: raw HTTP and an SDK.

curlPython SDKMessages & rolesSystem promptTemperature

Security & Optimization

3 tasks

Treat AI APIs like any other production dependency — cost and abuse matter.

Environment variablesRate limitingToken optimizationCaching

Model Routing

2 tasks

One endpoint, many models — choose per request.

OpenRouterCost vs capabilityFallback models
Milestone

You can make authenticated API calls from code, with keys in environment variables and sane rate limits in place.

04

3-4 days · Intermediate

💻 Local Models

Run open-source models on your own hardware. Gain privacy, reduce costs, and understand what is under the hood.

0/7tasks complete

What you will achieve

  • Run an open-weight model locally and chat with it offline.
  • Explain weights and quantization in your own words.
  • Know when local models beat hosted APIs and when they do not.

Tools to Know

4 tasks

The local-model ecosystem you will actually use.

OllamaHugging FaceDocker Model RunnerLlamaDeepSeekQwen

Core Concepts

3 tasks

Understand what is actually running on your machine.

Model weightsQuantizationGGUF formatFine-tuning
Milestone

You have a model running locally via Ollama and can explain why a quantized GGUF file fits on your machine.

05

4-5 days · Advanced

🧭 Context Engineering

Context is everything when working with large codebases. Learn how to manage what the AI knows and when.

0/8tasks complete

What you will achieve

  • Feed only the most relevant context to a model.
  • Distinguish short-term, long-term, and semantic memory.
  • Connect a model to a tool using MCP.

Context Management

2 tasks

More context is not better — the right context is.

Retrieval strategiesContext window limitsSummarization

Agent Memory Types

3 tasks

Agents need different kinds of memory for different jobs.

Short-term memoryLong-term memorySemantic memory

Model Context Protocol (MCP)

3 tasks

The open standard for connecting models to tools and data.

MCP architectureServers & clientsTools & resources
Milestone

You have built a small MCP integration and can articulate a retrieval strategy for a large codebase.

06

~1 week · Advanced

🔎 RAG & Knowledge Systems

Retrieval Augmented Generation lets your AI work with private organizational data that was never in its training set.

0/7tasks complete

What you will achieve

  • Explain embeddings and similarity search.
  • Stand up a vector database and ingest documents.
  • Wire retrieval into an LLM query end-to-end.

Prerequisites — learn these first

3 tasks

RAG stands on three ideas. Get these before building.

Vector databasesEmbeddingsSimilarity search

Building a RAG Pipeline

4 tasks

Put the pieces together and test on real data.

ChunkingIngestionRetrievalGrounding
Milestone

You have built a working RAG pipeline that answers questions over your own private documents.

07

5-7 days · Advanced

🤖 Agentic AI

Go beyond single-model calls. Learn to orchestrate multiple specialized agents that collaborate to solve complex problems.

0/6tasks complete

What you will achieve

  • Describe the roles agents play in a multi-agent system.
  • Implement hand-offs between agents.
  • Build and run a working multi-agent pipeline.

Multi-Agent Orchestration

3 tasks

Start small — two cooperating agents beat one giant prompt.

Planning agentsMemory agentsTask agentsHand-offs

Frameworks

3 tasks

Pick one framework and build something real with it.

CrewAILangChainLangGraphAutoGen
Milestone

You have built a 2-agent pipeline where one agent's output reliably drives the next.

08

Final Stretch · Capstone

🏁 End-to-End Projects

Reading and exercises are not enough. Cement everything by building real, integrated projects from scratch.

0/8tasks complete

What you will achieve

  • Integrate multiple techniques such as MCP, RAG, and agents into one project.
  • Work fluently inside an AI-assisted IDE.
  • Produce documented, portfolio-ready work.

Goal: Build 5+ Projects

3 tasks

Projects are where the learning becomes permanent.

MCPRAGAgentic orchestrationCursor IDE

Project Ideas to Start With

5 tasks

Pick one that excites you and ship it.

AI Kubernetes agentAI DevOps teamAI Ops monitoringPrivate knowledge chatbotMulti-agent code reviewer
Milestone

You have shipped 5+ documented projects, each combining at least two core techniques from this roadmap.

Portfolio builder

Capstone projects for AI Engineering

Use these to show real applied skill, not just course completion.

01Portfolio

Private Knowledge Chatbot

Build a RAG app that answers questions over your own documents with source citations.

02Agentic AI

Two-Agent Research Pipeline

Create researcher and writer agents with a reliable handoff and measurable output quality.

03MCP

Custom MCP Tool Server

Expose one useful tool to an AI client with authentication, logging, and clear permissions.

04Automation

AI Workflow Automation

Use n8n or a Python workflow to summarize, classify, route, or notify based on incoming data.

05Capstone

Multi-Agent Code Reviewer

Use agents to inspect pull requests, summarize risks, and propose tests or fixes.

06Ops

AI Ops Monitoring Assistant

Connect an AI agent to logs or metrics and have it summarize incidents with safe tool boundaries.