A free study toolkit for system design, coding interviews and deep learning from scratch. Five static pages that teach rather than list: every idea opens to what it is, a diagram, and an answer kept shut until you have tried.
Your progress is read from this browser's own storage and shown on the cards. Nothing is sent anywhere, and there is no account. Clearing this site's data resets it.
What is inside
System design guide
A framework of six steps for any design question: requirements, core entities, the API, data flow, the high-level design and deep dives, each with a time budget. Then 15 deep-dive patterns, from scaling reads to agent safety, holding 98 techniques; each technique has its own diagram, when to reach for it and its trade-off. Then 8 designs worked end to end: NotebookLM, Prior authorization validator, Email and calendar agent, Cloud coding agent, Voice AI agent platform, like Vapi or Retell, Enterprise virtual assistant, Ticketmaster and Kitaru. A design reads one step at a time; the arrow keys move between steps.
Algorithms and coding
The Start tab covers how to code out loud and what each operation costs. Then 14 patterns, from two pointers and sliding windows to graphs, dynamic programming and backtracking. Each opens to an example, a template, classic problems and their variations. The page holds 34 problems and 87 variations, each with a full statement, two worked examples, constraints and the code, shut until you have tried. The AI systems tab has 14 topics beyond LeetCode, such as RAG over a parser's output, an agent loop, a voice pipeline and fine-tuning with LoRA: the design in components, a skeleton for each, what to remember and what gets asked. Every code block runs and is tested.
Cheat sheet
The coding page on one screen: a card per pattern and topic. Click a card for its examples and variations; F goes full screen. It fits one screen on a laptop and scrolls on a phone. It is also one self-contained file, with its fonts and figures inside: download it and it works offline.
Baseline
A map of 15 fields of software engineering, from systems and distributed systems to inference, audio and speech, graphics and games. Each field opens to an overview, a diagram of how its parts connect, and 268 topics in all, each a short explainer: what it is, where you meet it in industry, the catch, and what to read for depth. Read a field's outline in five minutes before a call; open what you could not explain in a sentence.
Deep learning from scratch
23 stages in three parts. Foundations covers the maths and how audio becomes numbers. In the second part you build a small PyTorch-like framework on NumPy, module by module: tensors, autograd, optimisers, convolutions, attention and a transformer. The third reads a real speech system, from speaker identity to how a text-to-speech model turns a written description into a voice. Each stage has a goal, a diagram, what to build and an exit check with its answer. Python and NumPy; no GPU.
Every diagram can be explored. Click one and it opens full screen. Hover a part for a one-line note and click it for the explanation, with links to where it is taught; Walk through steps the flow one part at a time, and the arrow keys move.
Who it is for
Engineers preparing for system design and coding interviews, including roles where a model sits in the middle of the system. Students who know Python and want the common patterns named and practised. Anyone who wants to understand deep learning by building it rather than calling it.
How to study with it
- Coding first. Read the Start tab, then take one pattern a day: read its template, try a problem before opening the solution, and tick it once you can solve it cold.
- One design at a time. Read the framework once, then work a design step by step, saying each step aloud before you read it.
- Patterns when a deep dive needs them. Open the pattern a design names and read the techniques it lists.
- The cheat sheet for revision. Keep it open while you practise, and download it for the week before an interview.
- Deep learning alongside. A few hours a week, in order. Pass each stage's exit check before you move on.