HN Showcase
Back to browse
AI-native tech assessments (end of LeetCode)

AI-native tech assessments (end of LeetCode)

by vetted_so·May 9, 2026·2 points·5 comments
Visit ProjectView on HN

AI Analysis

●●SolidSolve My ProblemShip It

Finally, an assessment tool that admits AI exists instead of banning it.

Strengths
  • •Evaluates candidates on fixing race conditions in real codebases, not reversing linked lists.
  • •Provides granular metrics on how candidates collaborate with AI, not just final output.
  • •Simulates a realistic dev environment with terminal access and AI agents included.
Weaknesses
  • •Hard to scale manually graded assessments compared to automated LeetCode-style tests.
  • •Risk of candidates prompt-engineering their way through without deep technical understanding.
Category
Productivity
Target Audience

Engineering managers and technical recruiters

Similar To

HackerRank · Codility · Take-home assignments

Post Description

Hi HN!

We built openround.ai - an assessment platform to hire AI native engineers. On OpenRound, engineers build a project or solve a problem using AI. This replaces a usual coding round/take home assessment.

I believe coding assessments measure who is good at passing interviews, not who is actually a good engineer.

We want to flip that by showing how engineers actually work on a real problem.

2 interesting problems we solve -

1. Designing problems and environments that are not one shotted by AI.

2. Differentiating bw a software engineer who uses AI well and a pure vibecoder.

Would love for you folks to try it and hear any feedback that you have.

Also here to answer any questions about problems, evaluations, market insights, etc.

Similar Projects

SaaS●Mid

SharpSkill – I Gamified a LeetCode-like tool to crush Tech Interviews

SharpSkill pairs timed, mission-oriented validation tests and flashcards with per-technology progress tracking and recruiter-facing scoring — so candidates train under interview-like constraints while employers get a shortlist based on client-oriented metrics. The homepage communicates feature-focused thinking (real recent interview questions, position-specific tracks, validation tests) and the UX feels intentional, but this is a very crowded space and I didn’t see an obvious technical or pedagogical differentiator that would make me switch from existing tools.

SlickSolve My Problem
CocoZozo
103mo ago
AI/ML●●Solid

AIBenchy – Independent AI Leaderboard

The reasoning-score idea — a separate LLM judging chain-of-thought efficiency and penalizing token-churn or looping — is the clearest clever move here, and pairing that with cost-per-result and consistency metrics makes the leaderboard actually useful for product decisions. The table UI is clean and immediate, but the project needs more transparency on test corpora, judge prompts, and reproducibility before I’d trust swaps in production.

Big BrainSolve My ProblemSlick
XCSme
113mo ago