1. The curriculum is written by a working engineer, then peer-reviewed by a second.
No intern drafts, no AI-only drafts, no unnamed contributors. The cohort lead is a staff-or-above engineer with at least a decade of production experience, and every chapter is read in full by a second senior engineer on the editorial board. Turnaround from first draft to your inbox averages eleven days of editorial work.
2. Office hours are live, weekly, and capped.
Twenty-eight seats per cohort. Not two hundred, not five hundred. Weekly office hours run for ninety minutes with the lead instructor plus one rotating editor; every cohort also has an asynchronous written forum where questions are answered within a working day. The instructor-to-student ratio runs at roughly 1:14 after accounting for the editorial reviewers, teaching assistants, and the forum moderators.
3. You ship a capstone to a real repository by week eight.
Every cohort closes with a capstone project reviewed in a private pull request against a starter repository. Past capstones have shipped as merge commits into open-source projects: a Linux eBPF load-balancer, a Postgres extension for vector search, a deterministic replay harness for a Rust async runtime. The benchmark repository on GitHub now exceeds 48,000 stars from these submissions alone.
4. The in-house runtime benchmark lab runs your code on ARM, x86, and WASM.
Every performance claim made in a cohort is reproduced on our hardware before publication: three architectures, three operating systems, raw data attached. You do not have to take a benchmark at our word; you can reproduce it locally against the same commit hash.
5. This is continuing education, not a credential.
We do not issue certificates of completion and we do not partner with universities for accreditation. The work you ship — the capstone, the forum threads, the bench numbers — is the credential, and it links to your public profile if you choose to publish it. Median experience across the 2024 cohort was 9.4 years of professional engineering.