Schedulr

Problem

A greenfield build: a scheduler for tasks and meetings. Clone the starter repo, use your own editor and AI tools, and build under a 60-minute timer.

The spec

• A Task { id, duration, dependencies[], availability[] } where availability is a list of time windows.
• schedule(tasks) returns a conflict-free assignment when one exists: each task placed in a window, after all its dependencies complete.
• When a conflict makes a clean schedule impossible, report which tasks conflict rather than producing garbage.
• Greedy first-fit placement is fine — you must be able to say why and where it's not optimal.

Two patterns meet here: topological ordering for the dependency half, and greedy interval placement for the time half. Deciding how they compose is the problem.

What interviewers watch

• Entities modeled before logic (Task, Window, Schedule)?
• Dependency order computed separately from placement — or tangled into one loop?
• What happens when constraints conflict: explicit reporting, not silent overbooking?
• Are the ordering and availability tests written first?

Submission

Export your code and AI chat transcript, note tools and time, and submit for a verdict.

Deliverables

• A task model with dependencies, duration, and preferred slots
• A scheduler that respects dependency order and availability windows
• Conflict-free placement when possible; explicit conflict reporting otherwise
• Tests for dependency ordering, gaps, and overbooking