Preparing your design workspace

Loading the problem and your latest saved attempt…

Monorepo Build and Vehicle Software Rollout

Design an end-to-end pipeline that computes affected monorepo targets, builds and stores immutable artifacts, and safely promotes vehicle software through controlled rollout waves.

Problem statement

Design the path from a commit pushed to a large monorepo through delivery of software to vehicles. The build system represents targets as a dependency DAG, determines every transitively affected target, and executes builds in a valid topological order. Successful builds produce immutable artifacts in object storage.

A release is promoted through test hardware, test vehicles, canary vehicles, and then the broader fleet.

For this exercise, assume each promotion wave has explicit eligibility checks, artifact-integrity verification, health evaluation, and an auditable pause, continue, abort, or rollback decision before the next wave begins. A release manifest pins exact artifact digests and compatibility metadata.

Acknowledged build and rollout state survives retries and controller failover. Repository size, dependency edges, commit rate, build concurrency, artifact volume, fleet size, rollout duration, health thresholds, and reliability targets are candidate estimates, not fixed product facts.

Cover change impact, DAG scheduling, caching, artifact provenance, release assembly, device targeting, staged rollout, health signals, rollback, offline vehicles, concurrency, and observability.

What to build

  1. Given a commit and dependency-graph revision, compute the transitive affected-target closure and a valid dependency-aware build plan.
  2. Execute the plan with retry-safe task identity, reuse only compatible cached outputs, and publish integrity-verifiable immutable artifacts.
  3. Create an immutable release manifest that pins artifact digests, compatibility constraints, provenance, and rollout policy.
  4. Target eligible vehicles and promote the release through test hardware, test vehicles, canaries, and the broader fleet with health gates and auditable continue, pause, abort, or compatible-rollback decisions.

At scale

  • Candidates should estimate repository targets, dependency edges, commit rate, build fan-out, worker concurrency, cache hit rate, artifact size and retention, fleet size, and rollout duration.
  • For this exercise, build targets form an acyclic dependency graph at an immutable graph revision; a release cannot include artifacts built against incompatible graph or toolchain revisions.
  • For this exercise, artifacts and release manifests are immutable and content-addressed, and clients verify digests before installation.
  • For this exercise, vehicles may be offline for extended periods; rollout decisions and acknowledgements are durable and idempotent, and an offline vehicle resumes from current eligibility and policy rather than an expired command.
  • For this exercise, every wave has a minimum evidence window and explicit health thresholds before promotion; concurrent controllers cannot advance the same release twice.