About - LLMs Visual Card

I built this resource because preparing for an AI engineering interview requires rapid, structured revision. You likely already know bits and pieces of how the technology works. But when you need to connect the dots fast before a technical screen, rereading documentation takes too long. This site gives you a high-signal way to revise core concepts visually.

These visual cards break down the mechanics of large language models into one visual concept per page. Each card pairs a clear diagram with a short explanation of what the concept is and why it matters in production. The sequence maps the entire pipeline, covering tokenization, transformers, training, alignment, RAG, AI agents, and deployment.

How to use it

Treat the map as your curriculum for end-to-end study. If your interview is tomorrow, use it for last-minute revision to refresh your memory on specific architecture tradeoffs. Visual frameworks are much easier to recall than text blocks when you are answering questions under pressure.

The free visual library: 193 cards

This site hosts the core foundational cards completely for free. It provides a solid practical understanding of the system so you can discuss AI architecture with confidence.

The complete collection: 332 cards

The AI industry moves quickly. New research adds useful techniques and changes engineering trade-offs. That is why I keep this project alive.

The expanded collection on Gumroad contains 332 visual cards. I continuously add new cards as new agentic workflows and LLM breakthroughs emerge. When you get the complete collection, you get all future updates for free. Use it as a visual reference alongside current documentation and hands-on practice.

Sharing and attribution

The public card images use CC BY-NC-ND 4.0. Share unchanged images for noncommercial use, with attribution to LLMs Research, a link to our website and the source repository, and the license. For commercial reuse, contact the lab for permission.