Schedulers are a deeply fundamental part of much of the software that we use: operating systems, server daemons, async runtimes, browsers, inference engines. A good scheduler is invisible. The work gets done and you never notice. A poorly built or mismatched scheduler yields pauses, errors, unfairness, and inefficiency. Yet most engineers (and agents, for that matter) don't understand schedulers well.

So let's fix that! We'll build a working scheduler together, from a simple token bucket up to a weighted hierarchy of interruptible long-running tenants. We'll demystify what a scheduler decides, how it works, and how to tell a working one from a broken one.

You'll leave knowing what a quantum is, what a deficit is, how a few common types of schedulers work, and why fast approximate fairness beats slow exact fairness every time.

Schedulers are a deeply fundamental part of much of the software that we use: operating systems, server daemons, async runtimes, browsers, inference engines. A good scheduler is invisible. The work gets done and you never notice. A poorly built or mismatched scheduler yields pauses, errors, unfairness, and inefficiency. Yet most engineers (and agents, for that matter) don't understand schedulers well.

So let's fix that! We'll build a working scheduler together, from a simple token bucket up to a weighted hierarchy of interruptible long-running tenants. We'll demystify what a scheduler decides, how it works, and how to tell a working one from a broken one.

You'll leave knowing what a quantum is, what a deficit is, how a few common types of schedulers work, and why fast approximate fairness beats slow exact fairness every time.