Free · for engineers preparing for interviews
Learn the pattern once.
Don’t lose it by April.
Patterns, problems, system design and Python in one place — with spaced repetition so what you learn stays learned, and a Socratic AI mentor that reads your live code and never hands you the answer.
No card, no trial. An account just saves your progress and review schedule.
The pattern map
8 patternsEach pattern builds on the ones to its left. Hover or focus one to trace its path.
What’s inside
- patterns
- 8
- problems
- 24
- system design challenges
- 66
- Python lessons
- 88
- Python libraries
- 14
Five tracks
Everything the interview loop asks for.
Algorithms by pattern, system design on a real canvas, and the Python you'll be asked to write — each with its own path, all feeding one review queue.
Patterns
Learn each technique once, in prerequisite order — then keep it with spaced review before it fades.
- patterns
- 8
System Design
Drag load balancers, caches, queues and shards onto a canvas, then get a Socratic review of your design.
- challenges
- 66
- sections
- 12
Problems
Python and JavaScript run in your browser tab — tests, a complexity check, no submission queue.
- problems
- 24
Python
Beginner to expert in four levels, with an example you can run and edit on every lesson.
- lessons
- 88
Python Packages
NumPy to PyTorch, ordered into Data Engineering, Machine Learning and AI paths.
- libraries
- 14
- paths
- 3
The gap
Volume isn't the thing you're missing.
You can grind three hundred problems and still blank on a variation you solved in November. Recall is the bottleneck, and nothing in the category is built for it.
| Capability | Problem banks | Curated lists | MentorNode |
|---|---|---|---|
| Organized by pattern | — | ||
| Editorials that teach the why | — | ||
| Spaced repetition built in | — | — | |
| AI coaching on your live code | — | — | |
| Refuses to show you the answer | — | — |
What you get
Three things, done properly.
Not a feature list — these are the three that change whether you walk into the interview able to think.
Taught by pattern, not by problem number
Core patterns, each with an editorial that argues why this pattern applies here — the reasoning you have to reproduce under interview pressure, not a solution you can only recognize afterward.
Spaced repetition that tracks patterns
FSRS schedules each pattern on its own decay curve and surfaces a daily queue of what's about to fade. You review Sliding Window the week it starts slipping, not the week before the onsite.
A mentor that reads your live code
Stuck at line 14? It asks the question that unblocks you and refuses to give the answer. Spoiler-free is the constraint the whole thing is built around, not a setting.
How it works
The loop.
Three phases, repeated. The scheduler decides when you come back, so retention isn't a thing you have to remember to do.
Learn the pattern
Start from the shape of the problem — the invariant, the window, the state — before touching a single problem that uses it.
Review before it fades
FSRS puts each pattern on its own decay curve and picks the day. You clear the queue; the patterns stay loaded.
> review("sliding-window", 3)
> stability: 12.4d → 31.7d
> next_review: +31 daysGet unstuck without being told
Solve it in the browser — Python and JavaScript run locally in your tab, no submission queue. When you stall, the mentor asks the question that moves you, and refuses to write the line.
I’m getting an IndexError on line 14.
Look at the condition on your right pointer. What is its value the moment the inner loop stops shrinking the window?
Pick a track. Start today.
MentorNode is free to use. If it helps you land the job, you can chip in to keep it running — entirely optional.