HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88

Job-scoring AI gave the same resume wildly different grades — and commenters are screaming

TLDR: A popular new hiring tool gave the same resume wildly different scores, raising fears that job seekers could be rejected by pure luck. Commenters split between outrage over unfair grading and grim acceptance that companies may still use it because sorting huge piles of applications is so hard.

The internet has found its latest hiring horror story: HackerRank’s new open-source resume screener looked impressive at first, then immediately face-planted. One person ran the same exact resume through it again and again and got scores bouncing from 66 to 99. That means a candidate could be rejected or passed for the very same job based on what commenters are basically calling digital coin-flip energy. The most viral reaction wasn’t shock — it was exhausted recognition. A lot of people read this and went, wait… is this why no one gets callbacks anymore?

The comments are where the real fireworks start. One camp is furious that the scoring system gives huge points for unpaid side hustle stuff like open-source work and personal projects, with one commenter bluntly saying they don’t spend their free time doing more job-related labor. Another camp delivered the bleakest hot take of all: even a broken filter might be “good enough” if companies are drowning in resumes. That sparked the kind of discussion people hate because it feels a little too real. And then came the jokes: one commenter said the AI had basically mastered the oldest Human Resources trick in the book — throwing out half the pile because “we don’t need unlucky losers.” Ouch. Between the chaos, the sarcasm, and the gallows humor, the mood is clear: people aren’t just worried this tool is unfair — they think it’s hilariously, terrifyingly on-brand for modern hiring. For anyone job-hunting, this story hit like a jump scare.

Key Points

  • The author tested HackerRank’s open-source ATS on the same resume across 100 runs and observed total scores ranging from 66 to 99.
  • The tool parses a PDF resume, uses six LLM calls to extract structured sections, adds GitHub repository context, and then performs a final LLM-based grading step.
  • The scoring rubric assigns 35 points to open-source work, 30 to personal projects, 25 to work experience, and 10 to technical skills, plus up to 20 bonus points.
  • Technical skills scoring was highly consistent in the author’s tests, while project scoring varied substantially across runs even at low temperature settings and when using a different model.
  • The article found experience scoring gave full marks to both the author’s current resume and an older resume with one internship, and traced that behavior to a very short prompt lacking a detailed rubric.

Hottest takes

"35% chance of elevating a technical individual to the next stage with no effort?" — jerrythegerbil
"I don't contribute to open source or have personal projects because I don't spend my free time doing what I ..." — dc3k
"we don't need unlucky losers" — cyberax
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.