How I structured a technical assessment for hiring remote developers (and what actually worked)

#typescript

Running a 3-person startup means every engineering hire is high-stakes. Here's the assessment framework we landed on after several failed attempts.

What didn't work

Live coding in a video call — candidates froze up, output didn't reflect their actual ability. Senior devs especially hated being watched.

Full take-home projects (3–5 hours) — good candidates dropped out because their time is valuable and they had options.

Just asking about past projects in an interview — too easy to embellish, no signal on actual implementation thinking.

What worked: a focused 90-minute async task

The format we converged on:

A real, scoped problem from our actual codebase context (not a LeetCode puzzle)
90-minute time limit, explicitly communicated upfront
Asynchronous — they pick a 2-hour window, we review async
Evaluated on reasoning, not perfection

We send a single file: a TypeScript module with a bug, a missing feature, and some intentionally messy code. The task:

Fix the bug (tests are provided, they should pass)
Add the missing feature (spec is included)
Optionally: clean up anything that bothers you

What we evaluate

Bug fix correctness: Do they actually read the failing test?
Feature implementation: Do they handle edge cases?
Code style choices: Consistent, readable, no over-engineering
Optional cleanup: Shows initiative and code taste
Time management: Did they scope appropriately in 90 min?

What we explicitly don't evaluate

Whether they used the same patterns we use (we want their instincts)
Speed (90 minutes is enough to see real work, not a rush)
Perfect syntax (they can Google things)

The follow-up conversation

After the assessment, we do a 30-minute async written Q&A (not a call). We ask:

"Walk me through your fix for the bug"
"What would you change if you had another hour?"
"What bothered you most about the original code?"

The written format gives introverted engineers a chance to shine and reveals how well they communicate technical decisions in text — which matters enormously for async remote teams.

Results

We've used this format across ~40 hires. Key observations:

Drop-off rate is much lower than take-homes (candidates respect the 90-min scope)
Strong signal on seniority without a 5-round process
Candidates who do well here almost always perform well in the role

Happy to share the actual task template if there's interest — drop a comment.