Live data from Hacker News

Viewing profile — gurachek

gurachek

HN member
Joined
Wed, Jun 12, 2019, 6:23 PM UTC
HN karma
21
Public activity
26 items

About gurachek

having 24/7 fun with agents, very serious backend engineer email - gurachek at proton dot me

Recent public activity

  1. comment
    Comment #48880667

    I think engineers are misreading part of the 2026 job market. Some roles are not being eliminated. They are never being created. That distinction matters because layoffs recover cy…

  2. story
  3. story
  4. comment
    Comment #47562954

    Yeah, I think so. So far, Claude Opus is the only model I found that doesn't fold under the minimal pressure and can push back, but still - push just a little bit harder and it's b…

  5. story
  6. comment
    Comment #47555406

    I had exactly this between two LLMs in my project. An evaluator model that was supposed to grade a coaching model's work. Except it could see the coach's notes, so it just... agree…

  7. comment
    Comment #47550799

    The examples in the article are all big scary wipes, But I think the more common damage is way smaller and harder to notice. I've been using claude code daily for months and the wo…

  8. comment
  9. comment
    Comment #47526008

    The compounding booboos bit is the key insight here. Humans are a bottleneck and that bottleneck is actually load-bearing. You feel the pain of bad decisions slowly enough to cours…

  10. comment
    Comment #47525149

    Raw audio is a cool idea! I've seen a similar approach in other domains, "keep the source of truth accessible so you can verify the AI output against it". I wouldn't go as far as "…

  11. comment
    Comment #47525080

    The float comparison slider is great. One thing from practical experience - the quality gap between model sizes shows up in a way benchmarks don't capture. I have a system where a …

  12. comment
  13. comment
    Comment #47524530

    The union rep gets it - people improvise when you cut their tools and then threaten discipline for improvising. That memo is how you make staff hide things instead of asking for he…

  14. comment
    Comment #47479830

    Yeeah, LLMs are the perfect procrastination tool because they feel productive. You're "learning", you're "exploring", you're having this great conversation about the problem. And t…

  15. comment
  16. comment
    Comment #47479059

    Your "no compiler" rule on day 3 taught you more than the LLM did. The LLM made concepts click. But the binary search vanishing under interview stress proves that understanding som…

  17. comment
    Comment #47283249

    I wrote this after going through Fabric HQ's data. The 61% stat surprised me - the cheating actually works, at least for now. Happy to discuss the implications.

  18. story
  19. comment
    Comment #47233051

    Forgot to include: answers not stored. Here is DB schema(content hash is to save money on requests with the same answer): | # | column_name | data_type | |----|--------------|-----…

  20. comment
    Comment #47232851

    I built this because it's easier for me to treat interview & job search as a math game, and math needs numbers(lol?). The original component works with voice, because engineers del…

  21. story
  22. story
    Show HN: Prepto.tech – Interviews are simple with AI (no cheating)

    I've been through 50+ technical interviews as both a candidate and an interviewer. The current system is broken: we memorize algorithms we'll never use while real job skills go unt…

  23. comment
  24. story
  25. comment
    Comment #31531891

    Hey guys! It's a concept at the moment. But still, what do you think about such a tool if it's real? Should I invest more time building such stuff?