An interview scorecard for a nearshore engineer

A useful nearshore engineer interview tests the work the person will actually do inside your team. Start with a representative task, score observable behavior, and keep working-hour availability separate from technical capability. A confident conversation alone cannot establish readiness to own a production change.

For Costa Rica and US teams, written communication deserves a place in the assessment. The relevant question is whether colleagues can act on a handoff, not whether a candidate sounds like the interviewer.

Define the assignment before the questions

Write down the role’s first meaningful responsibility. Maintaining a mature API calls for a different interview from building a new interface. Choose a small, fictional repository or scenario that resembles the responsibility without requiring unpaid production work.

Explain the exercise format, available tools, and expected artifacts in advance. If documentation or AI assistance is allowed, use the same policy for every candidate. Ask people to explain and verify any generated code rather than treating tool use as proof of competence or incompetence.

The US Office of Personnel Management’s structured interview guidance describes using consistent questions and rating standards. The engineering scorecard below is a proposed adaptation, not an OPM technical hiring standard.

Use anchors instead of adjectives

Avoid scores labeled only “good” or “senior.” Record what happened and how much prompting was necessary.

Dimension	Needs substantial support	Ready for independent work
Diagnosis	Changes code before reproducing the issue	Reproduces, narrows the cause, checks an assumption
Implementation	Fixes the visible symptom only	Fits the local design and handles relevant boundaries
Verification	Says it should work	Demonstrates a regression check and its limits
Judgment	Adds complexity without a reason	Explains a tradeoff and a simpler alternative
Collaboration	Leaves an unexplained patch	Writes an actionable summary and names open questions

Add an intermediate anchor for someone who reaches the independent behavior after a specific prompt. Keep “not observed” separate from a low score. Missing evidence should trigger a focused follow-up, not an invented rating.

Hypothetical example: a duplicate background job

Imagine a candidate receives a small service where retrying a request schedules the same job twice. This is a hypothetical interview exercise.

Candidate A quickly adds a delay and declares the defect fixed. Candidate B asks how requests are identified, reproduces the duplication, proposes a bounded fix, and adds a test that fails before the change. Neither typing speed nor verbosity settles the assessment.

In the written handoff, Candidate B notes that the exercise does not cover concurrent requests and describes the additional check needed. That is useful evidence of judgment. The interviewer should still test the gap rather than assume concurrency expertise from a sentence.

Decide what cannot be averaged away

Choose essential dimensions before interviewing. For a role expected to ship independently, weak verification should not disappear inside an attractive overall average. If the role includes mentorship and review support, say so and assess against that scope.

Have interviewers record evidence independently before comparing notes. Discuss disagreements using the candidate’s actions, and verify availability through a separate conversation about actual overlap windows, including US seasonal schedule changes.

Draft one representative exercise and its scoring anchors before requesting candidates for staff augmentation. It gives the hiring discussion something firmer than a list of frameworks.