I applied online. The process took 2 weeks.

The process began with a 30-minute screening containing technical questions. Following a positive evaluation, the recruiter scheduled an additional 30-minute preparation call to review guidelines and expectations for the subsequent technical rounds. While recruitment logistics and communication were prompt and well-handled, the technical coding round significantly deviated from what was prepared for and framed during briefing. The 30-minute live coding session was presented as a collaborative, interactive exercise, but the reality was uncalibrated:

• Lack of Collaboration: Rather than functioning as a paired, interactive discussion, the dynamic felt like waiting for failure points under an unrealistic 30-minute delivery requirement for fully functional code.
• Ignored Trade-offs: To handle the strict time constraints, I explicitly proposed using AI scaffolding as a baseline to walk through architectural design, code review, and iterative refinement. I communicated repeatedly that the output was not intended to be production-ready code out of the box.

The "No-Win" Dynamic & Arbitrary Failure Angles. The tight constraint made genuine success impossible because any chosen approach could be arbitrarily penalized based on personal interviewer preference:

• Option 1 (What I chose): Deliver a working, thread-safe solution with known inefficiencies that were explicitly highlighted and critiqued line-by-line during the call.
• Option 2 (Small, clean stub): Deliver high-quality, incomplete code, leaving the candidate open to criticism for moving too slowly and producing nothing functional.
• Option 3 (TDD approach): Grow the system incrementally from unit tests, risking the critique that domain modeling was skipped up front.
• Option 4 (Domain modeling first): Focus on upfront design and schema, risking failure for not producing working code in a live coding round.

Passive Observation. Attempts to facilitate meaningful dialogue around these trade-offs were met with passive silence, leaving the evaluation entirely dependent on the interviewer's unspoken bias.

Post-Interview Feedback & Factual Discrepancies. What prompted this review was not just the constrained format, but the fact that the post-interview feedback was objectively incorrect:

• Distorted Evaluation: The feedback warned to "critically evaluate AI output," directly ignoring that this was the exact methodology I explicitly stated, demonstrated, and walked through during the session.
• Dismissal of Context: Intentional architectural trade-offs that were clearly communicated were treated as careless mistakes rather than conscious adaptations to severe time limits.

Just to reiterate, I am not questioning the decision. I am providing this feedback because the feedback itself was factually inaccurate, and because the process created a no-win scenario despite presenting itself in a way that led candidates to believe it would be fair and that the stated expectations could be relied upon.

Differences between sharing and partitioning