Top Tweets for #TechInterview

Technical interviews are completely outdated, and I’ve overhauled the process at my company I’m an engineering manager at a large corporation, and hiring is part of my job. I’ve never liked the way tech companies run interviews, including Google, Microsoft, Amazon, and the rest. It’s a stressful, needlessly complicated process that doesn’t test what actually matters. These days, companies either require in-person interviews or ban AI entirely. Both approaches are wrong What’s wrong with interviews today 1. The recruiter screening stage is often useless. People have learned how to lie about their experience, and no recruiter can tell the difference. 2. Interviewers grill candidates on knowledge that isn’t what the company actually needs. Understanding system design matters more than solving hardcore algorithm problems. Knowing how to get useful information out of AI and apply it correctly matters more than knowing the documentation for a particular tech stack 3. Interviews are too narrowly focused. Nobody looks at how candidates broaden their understanding to build projects 4. We need to stop assessing people solely on their ability to follow instructions. An engineer’s role is no longer to execute tasks, but to organize the work. 5. The process is demeaning. Interviewers put themselves above candidates and give them unnecessarily difficult, poorly designed problems. Traditional interviews measure stress tolerance, not skills 6. There are no consistent criteria. Every interviewer judges candidates by their own standards and their mood that day, and the final decision comes down to gut feeling. The same candidate passes with one interviewer and fails with another. It’s a lottery What I changed 1. Recruiters now have to send me the resumes of everyone they’ve screened - both those they’ve advanced and those they’ve rejected with a brief explanation of each decision. I review some of the rejected candidates and invite them to interview when I think they deserve a closer look. There’s real talent in that group 2. I’ve removed time limits entirely. They’re one of the biggest sources of stress. But there’s an important caveat: if a candidate is doing very poorly or very well, we stop the interview rather than spend more time unnecessarily 3. I allow unrestricted use of AI, chatbots, agents, Google, documentation, and any other information available online. Candidates need to show how they work with publicly available information and apply it in unfamiliar situations. Can they take initiative and use those resources to go further and do better? 4. No more stupid academic algorithm puzzles or other textbook nonsense. The candidate chooses a small project and builds it live. I’m not looking at the code; I’m looking at how they think about the product: what they’re building, why, and how it will evolve over time. 5. A system design round is now mandatory for everyone, from product managers to backend developers. But the task is tailored to their role. There’s no point asking a product manager about quorums in Cassandra. 6. I’m a teammate, not an interviewer. If someone gets lost or doesn’t know something, I help and offer hints. I don’t penalize mistakes or gaps in knowledge, and I give candidates another chance through follow-up questions. How someone uses those hints tells me more about them than any correct answer. The results 1. The process is much better at identifying motivated people with genuine drive. Someone may have gaps in their knowledge, but they have the potential to move in the right direction, find new information, and apply it effectively 2. People stop feeling nervous and do a better job of showing what they know. You find genuinely valuable people, rather than extroverts who are good at selling themselves 3. Emotional intelligence, teamwork, and other personal qualities become much easier to assess 4. You find people who can actually work in modern software development not just pretend they can. 5. But I arrived at this process on my own, once I realized that what matters isn’t how much you know, but how much knowledge you can tap into. I think this principle should apply to every field, including schools and universities