Most preparation sites hand you the pattern, the approach and the solution on one page and call it studying. You leave knowing you could have solved it. Forge makes you sit with the problem first — hints unlock on a clock, the pattern stays hidden, and what you reveal is recorded against the attempt.
Every problem hides its pattern, its approach and its solution behind gates that open on a timer you choose. Drill opens them early, Study opens everything, Mock opens nothing until the budget is spent. The pattern label is treated as a spoiler, because naming the pattern is most of the answer.
A bank that labels every problem "DP — Knapsack" has already told you what to do. Here the list shows tier and status; you meet the problem before you meet its category, which is the only way the practice resembles the interview.
91 labs across 23 tracks, in their own public repository — perf counters, cache behaviour, codegen, concurrency, syscalls, AArch64 and SVE. Each opens as a guided session: what it teaches, the build line, then 171 steps with the command to copy and the output that step produced on the machine that wrote the lab, so you compare against a real number instead of guessing. It is finished when you can write what it showed on YOUR machine and what surprised you.
Not by company folklore. 3 roles, each with the day it actually involves and its competencies sorted into foundation, working and expert. Problems, labs, guides and recall questions hang off those competencies, and picking one re-sorts every surface.
121 mechanism questions across C++ internals, systems, architecture, compilers, probability and microstructure — one at a time, nothing else on screen: commit out loud, reveal, then see the follow-up where most people come apart. Plus 19 estimation questions that make you write your chain of assumptions before showing you a worked one, 32 C++ concepts with the mechanism and the follow-ups, and 33 patterns each linked to the problems that need it.
Everything works signed out, saved in your browser. Sign in with Google and it follows you between devices, merged rather than overwritten, so a session on your laptop never erases one from your phone.
Pick one and every surface re-sorts around it: which problems matter, which labs to run, which guides to read, which questions you must answer without preparation.
You own a stage of a pipeline that turns text into machine code, inside a codebase of millions of lines that thousands of people depend on being correct. Most days are not writing new optimizations. They are reducing a bug report to twelve lines, bisecting which pass broke it, reading IR before and after, writing the FileCheck test that would have caught it, and defending the patch in review.
Someone hands you a program that is too slow and no idea why. Your value is that you do not guess. You build a repeatable measurement, find the bottleneck with evidence, name the mechanism, change one thing, and prove the change with numbers a sceptic would accept. Half the job is refusing to accept a plausible story without a counter behind it.
A packet arrives and a decision has to leave the box before anyone else's. The budget is hundreds of nanoseconds, so nothing on the hot path may allocate, block, lock, or call the kernel. You spend your days shaving a path you have instrumented at every stage, arguing about data layout, and being certain about memory ordering because the bug you cannot reproduce costs money every day it survives. The loop also asks you to reason about probability out loud and to know enough market structure to understand what the code is for.
Next.js and React, with Supabase for Google sign-in and Postgres when you want it — and fully functional without it. Two repositories: the site, and the labs you run on your own machine. Clone either, npm install, npm run dev.
27 guides and 91 labs, every one its own page. Browse without signing in.