The setting

Juniper Freight gives small trucking fleets live shipment tracking. About 30,000 trucks report positions to an AWS-hosted platform run by a four-person infrastructure team, and the product is growing faster than the architecture.

Every mission on this path happens at the same company, so context carries over the way it does in a real job: the data you cleaned in mission two is the data the finance lead questions in mission four.

The missions

1. architecture sketchstarter

Sketch the proof-of-delivery photo upload architecture

Juniper Freight is adding proof-of-delivery photos: drivers photograph the signed paperwork at drop-off and shippers view it in the app. Legal requires seven years of retention, and drivers work on patchy mobile networks at loading docks. Product wants a design this week.

You deliver: A written architecture sketch covering the upload path, storage and lifecycle design, access control, and the sizing math.

Scored on: Upload path, Storage and privacy, Lifecycle math, Traceable flow.

Working from: feature_notes.md, current_stack.md.

2. service choicestarter

Pick the compute model for geofence alerts

Fleet managers want an alert within five seconds when a truck enters or leaves a customer geofence. Three build options are on the table and the team is split. You have been asked to make the call and defend it in writing.

You deliver: A decision memo recommending one compute model, scored against all four constraints with the traffic data applied.

Scored on: Tested against all constraints, Uses the traffic profile, Honest tradeoffs, Clear decision.

Working from: options_memo.md, traffic_profile.csv.

3. cost checkcore

Explain the March cloud bill jump to the CFO

The March AWS bill came in about 2,200 dollars higher than February and the CFO wants an explanation plus a savings plan before approving next quarter's budget. You have the cost export and the team's standup notes. Guesswork will not survive the meeting.

You deliver: A cost review memo that explains the March increase, quantifies untagged spend, and ranks three savings actions with monthly estimates.

Scored on: Finds the spike, Untagged spend surfaced, Savings ranked with numbers, Numbers accurate.

Working from: aws_costs_q1.csv, platform_notes.md.

4. resiliencecore

Find the single points of failure before Halloran asks

Halloran Trucking, 18 percent of Juniper's revenue, sent an uptime questionnaire asking for 99.9 percent availability and written answers on failure scenarios. Before anyone signs anything, you need an honest map of what actually breaks and takes the platform with it.

You deliver: A resilience review with named single points of failure, per-scenario blast radius, prioritized fixes, and a draft answer for the customer questionnaire.

Scored on: Finds the single points of failure, Blast radius reasoning, Prioritized fixes, Honest customer answer.

Working from: architecture_notes.md.

5. handoffstretch

Hand the Juniper platform to your successor

You are moving to the data team at the end of the month. Dara, your successor, arrives Monday with strong general AWS skills and zero context on Juniper. Everything living in your head has to land on paper: the quarter's half-finished work, the risks, and what must not be dropped.

You deliver: A handoff document with a platform overview, per-workstream status and next actions, a reviewed risk register, and a dated first-two-weeks checklist.

Scored on: In-flight work accurate, Risk ownership, Successor-ready checklist, Self-contained.

Working from: quarter_recap.md, open_risks.csv.

How the scoring works

Each deliverable is graded against the rubric written for that mission. Separately, every mission on every path is graded on how you used AI, against the same four criteria:

  • Understood the task. The learner framed the goal for the assistant clearly instead of pasting the brief and hoping.
  • Grounded in the material. The learner directed the assistant into the provided files and based the work on them, not on invented facts.
  • Verified the output. The learner checked claims, numbers, or coverage against the source material before submitting.
  • Iterated with judgment. The learner refined weak parts of the draft with specific follow-ups rather than accepting the first answer.

Both scores, with the work behind them, go on your proof profile. That is what makes a claim like "I can use AI for cloud engineering" something an employer can check.