Roadmaps

From Foundations to Production AI.

AI isn’t learned in a day, but it can be mastered through structure. That’s why we built this page: to turn a busy feed into a clear roadmap.

Whether you’re looking to build a second brain, understand the foundations of AI agents, or architect enterprise-grade systems, you can find your starting point here without digging through the archives.

We’ve organized our content in categories so you can explore at your own pace:

By Level: Beginner, Intermediate, Advanced.
By Collections: Concepts, Case Studies, Projects, Playbooks, etc.
By Series: End-to-End Roadmaps.
By Topics: MLOps, LLMs, AI Agents, RAG, etc.

Level

Beginner | Intermediate | Advanced

Collections

Concepts: One idea or mental model, explained until it clicks.

Case Studies: Real systems in practice: what worked, what broke, and why.

Projects: Build-alongs that end with a running system and a repo you can fork.

Playbooks: Tool-agnostic methods for doing one thing well.

Guides: Hands-on walkthroughs of one specific tool, framework, or SDK.

Roadmaps: What to learn or read, and in what order.

Series

AI Agents Foundations: A framework-free, first-principles roadmap to learn to build AI agents from scratch, covering everything from tools to planning and memory.
AI Evals Foundations: Learn how to integrate AI evals into your AI app to address your specific business problems and track and improve your product.
Designing Enterprise MCP Systems: A practical guide to architecting modular, production-grade AI systems using the Model Context Protocol (MCP) to build enterprise AI automation apps.
Second Brain AI Assistant: An end-to-end guide to engineering a production-ready agentic RAG system that connects LLMs to your personal knowledge base using MLOps best practices.
PhiloAgents: A hands-on series to build an AI-powered game simulation engine that brings historical philosophers to life. Learn to impersonate historical figures using LangGraph, agentic RAG, and long-term memory.
H&M Real-Time Personalized Recommender: A roadmap for architecting real-time recommenders using the FTI architecture and Polars. Covers 4-stage design, two-tower neural networks, and MLOps workflows with the Hopsworks AI Lakehouse.

Topics

Agentic Engineering
Use agents to ship real work: coding setups, autonomous loops, and automation pipelines that multiply one engineer’s output.
AI Agents
Agents that survive production: planning, tool use, MCP, multi-agent orchestration, real deployments.
↳ Agent & Workflow Patterns → Pick the right shape first: workflows vs agents, the autonomy slider, and guides against overengineering.
↳ Agent Memory & The Context Layer → Give your agent a past: knowledge graphs, GraphRAG, LLM wikis, and a context layer that follows you across tools.
↳ Context Engineering → Control what enters the context window and what comes out: prompting, structured outputs, recursive exploration.
↳ Harness Engineering → The runtime scaffolding around the model: the agent loop, tool-calling, sandboxing, permissions.
AI Engineering Career
What to learn, in what order, and what to skip: roadmaps, books, and hard-won lessons from the field.
AI Evals
Prove your AI works: benchmarks, eval suites, LLM judges, error analysis, and evaluation-driven development.
AI Systems
End-to-end blueprints for AI products: architecture, deployment, scaling, cost.
LLM Engineering
How models work and how to bend them to your domain: fine-tuning, datasets, embeddings, inference internals.
MLOps
What keeps a model alive: pipelines, CI/CD, infrastructure as code, observability, and serving.
↳ LLMOps
RAG & Retrieval
Get answers from your data: retrieval architectures, vector databases, semantic search, and prove the retrieval worked.
↳ Streaming & Ingestion → streaming, CDC, and crawling/ETL feeding your vector DB.
Recommender Systems
TikTok-grade personalization end-to-end: two-tower models, the 4-stage architecture, real-time serving.
Strategic Programming
Software fundamentals and design principles that last the AI era.

Resource Library

If you’re looking to go deeper with more structured guides, here’s where to look next. While the weekly content is great for staying sharp, if you’re ready to build a complete system from scratch without piecing together different articles, I’ve compiled the best of what I know into a few digital products.

Unlike the weekly posts, these include full codebases, video walkthroughs, and Q&A support to help you go from a blank IDE to a deployed system.

The LLM Engineer’s Handbook: A framework for building LLM and RAG apps.
Agent Engineering Course: The end-to-end blueprint for designing, testing, and deploying autonomous agents.
Full Course Catalog: Write real code. Ship AI that actually works.

Not sure what to pick? I also have a 6-day free email course on the critical design mistakes that silently break agentic systems. It boils down 2+ years of production experience into a simple mental model for building reliable agents that actually scale.