The setting

Harborline is a freight forwarding platform that mid-size importers use to book ocean container shipments. You are a Java backend engineer on the booking services team, which runs a Spring-based monolith splitting into services.

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. OOP designstarter

Review the shipment domain model rewrite at Harborline

A contractor rewrote Harborline's shipment domain model before rolling off, and the team inherited it without a review. Two production oddities already trace back to it, including duplicate shipments appearing in a HashSet-based dedupe.

You deliver: A design review of the shipment domain model with a proposed restructuring.

Scored on: Finds the design flaws, Explains the HashSet oddity, Sound restructuring, Grounded in principles.

Working from: domain_model.md, oddities.md.

2. REST servicestarter

Fix the booking API contract before partners integrate

Harborline is opening its booking API to two partner forwarders next month, and the draft contract went out for comment. One partner's tech lead replied with a single sentence: this does not look like REST to us. Your manager wants a real review before the kickoff call.

You deliver: An API contract review with a corrected endpoint-by-endpoint contract sketch.

Scored on: Finds the contract flaws, Partner-facing consequences, Corrected contract sketch, Workable pagination.

Working from: api_draft.md.

3. data accesscore

Hunt down the query storm behind the bookings page

The partner bookings page takes 4.2 seconds to render and the database team traced a burst of near-identical queries to it. Separately, a security review is due next month and the data access layer has never been audited. Both trails lead to the same repository class.

You deliver: A data access review quantifying the N+1 pattern and fixing the injection and resource-handling defects.

Scored on: Quantifies the N+1, Injection vector found, Connection leak found, Impact estimated.

Working from: booking_repository.md, query_log.md.

4. testingcore

Pin down the tariff calculator with a JUnit plan

Harborline's tariff calculator prices every booking, and finance found invoices that disagree with the published tariff by a few hundred dollars or a stray cent. The calculator has zero tests. Before anyone patches it, the team wants a test plan that captures the tariff rules exactly and shows precisely where the code...

You deliver: A JUnit-style test plan for the tariff calculator with exact expected charges and documented deviations of the current code.

Scored on: Tier boundary case, Money arithmetic case, Tariff rule coverage, Test plan quality.

Working from: tariff_rules.md, tariff_code.md.

5. performance tradeoffstretch

Choose between the rate cache and the nightly precompute

Quote latency is Harborline's top partner complaint: the quotes endpoint fans out to six carrier APIs and its p99 has crept past 2.4 seconds. Two competing fixes are on the table, a Redis rate cache and a nightly precompute of popular routes, and the team is split. Engineering leadership asked for one written...

You deliver: A recommendation memo choosing one latency fix with quantified impact, explicit risk handling, and a launch metric.

Scored on: Grounded in the data, Staleness handled, Quantified impact, Failure modes and metric.

Working from: endpoint_latency.csv, options_memo.md.

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 java backend" something an employer can check.