返回资讯中心
外部精选
管理/文化
#开发者文化#职业成长#qa#developer-story#developer-hiring#se-tech

How do you hire a human? Q&A with Tom Harvey, Stack Overflow technical recruiter

Everybody's resume is perfect now, so how do you find the perfect teammate?

Stack Overflow BlogRyan Donovan2026/11/12 · 5 天前30 分钟阅读

以下正文同步自 Stack Overflow Blog,版权归原站所有,已转换为易读排版。

November 12, 2026How do you hire a human? Q&A with Tom Harvey, Stack Overflow technical recruiter

Everybody's resume is perfect now, so how do you find the perfect teammate?

Hiring and getting hired as a developer isn’t what it used to be. Jobs feel scarce, and candidates plentiful, as every open role gets hundreds of applications. Both sides use AI to create and review candidates, sometimes in defensive ways to spot the other’s AI use. To see what hiring human developers looks like in this antagonistic hiring environment, I chatted with Stack Overflow’s own Senior Manager, Talent & Culture, Tom Harvey.

Q: Can you give an overview of what a tech recruiting process looks like, either starting from when the resume comes in or before when the job posting goes up?

A: Initially, you work quite closely with your hiring manager to establish the must-haves, the nice-to-haves, the things that you can't necessarily pass into a job description that will be helpful in the search.

If we look at Stack Overflow, we're less tied to how good are you at language X, but more interested in how good are you at talking about what you've built—trade-offs, how you work with other people in the team, and how coachable you are. I think a lot of the problems that we run into aren't necessarily replicated in lots of other businesses or companies. So, we tend to be more open-minded there.

That side of things happens before you even start talking to candidates so that you are informed enough to sell the job. That's becoming more and more important as we're starting to engage passive candidates in the market rather than rely entirely on applications. You want to have confidence in how you're assessing if someone would actually enjoy doing this job here.

The role of an engineer becomes less and less about writing code and more and more about how they participate in discussions around direction, strategy, and and all these things that would have been handed off before. You want to establish—particularly as you're talking to candidates—that they can or are willing to learn how to do that.

Q: It sounds like the technical recruiting process is harder to put qualitative markers on, right?

A: Yeah, but they're becoming the most important markers. Strong proficiency in a programming language doesn't carry the same weight that it used to, because anyone can show they've got a strong proficiency in something now with vibe coding. The qualitative markers—what are derogatorily called soft skills—are now becoming the core skills.

It's becoming harder in some ways to coach talent towards the importance of that if they're looking for a new job, because I do think that companies know that that's what they need. I don't know if that's quite from the job market where I think people still feel like the measure of success is "I'm good at X" versus "Here's a story of something I've done before.”.

Q: You could assess a candidate in the past on the more quantitative stuff, like four years of language X, you pass these coding tests, whatever.

A: Yeah. And and the tests are irrelevant now. A LeetCode test doesn't work anymore because the LLMs are making it easy to get there. We're not necessarily testing skills at pair programming in the same way now.

I know some companies do, "Let's just throw you some buggy code that AI churned out and let's fix it." That's becoming much more of a desired skill at the technical proficiency stage. But again, where's the way of displaying that in the market at the moment?

Q: I think that's hard because you have hundreds, thousands of applications coming in for every job role you post. How do you sort those out?

A: The biggest problem is the cruelest irony: everyone looks amazing now. It used to be that you could have a cursory glance at a resume and go, "No, I don't quite think that's right because it's too wordy for what I'm looking for," or there's too much or not enough detail. Now, everyone looks like Han Solo in Star Wars.

So I think the resume doesn't necessarily paint the picture reliably enough anymore. Candidates are in danger of overselling their experience and then that falling apart when they get to an interview stage.

What I need more is qualitative insights into experience—and I mean Experience with a capital E. It's not just "I've worked at company X for so many years"; it's "I did this specific thing, and this is what turned me into the developer I am today. These four events, this production code that stopped working, or this bug that destroyed everything, or these are the things that taught me what to do better next time." Those are much more interesting, engaging, and more reliable sources of experience than what we used to rely on before.

Q: I think the issue is still like, how do you get to that point? Because you have to filter out these resumes. You can't talk to everybody—unless you use an AI screener, I guess.

A: Which I don't want to do. I don't want to put myself out of a job, and also I don't think AI screeners are a good candidate experience either, you know? I still want to talk to a person who is excited and engaged, informed enough before that call about who they are, what they've done, and has come with interesting questions about that.

It's like, I'd love to see the trailer to your film, and then I'd like to watch the full thing. It's that buildup to the conversation that leaves me with good questions to ask that are more interesting for you, the candidate, as well.

At the screening stage, most modern ATSs (Applicant Tracking System) are still optimized for keyword filtering. They're still optimized towards very generic things like the right location. Even then, there's no real way of knowing now. There's fraud detection that are there to help you with addresses, but because you're going into that at a level of distrust, I don't think that makes for a great experience for me as the recruiter, or the hiring managers, or them as the candidate either. That distrust is leading to a difficult start to what should be a positive fact-finding exercise.

Q: Let's say somehow you have made a subjective judgment to get somebody to where they talk to you, and technical recruiter screens were always the first one where it's like, are you at all a good fit? Do you fall apart immediately? How much knowledge of what the candidate has to do do you have to bring to that conversation?

A: As in what they have to do when they do the job?

Q: Yeah. How much tech do you have to talk with them, or are you just like, can you hold a conversation?

A: That has changed rapidly in the last six months.

It used to be: can you hold a conversation, and can you read the audience effectively enough to go, "Okay, you're in that position, so I'll edit my answer. I'm going to assume you don't know very much.” Which is fine.

Now, the expectation of me as a tech recruiter is that I have to ask much more technical questions. I myself have to be much more technical. I have to understand what the team does, what a Monday on that team looks like, what you are going to be doing, and what I know about where the business is going to be six months from now so I can talk you through that to establish if this is the place you want to be.

But I also think that, ironically, because the expectation is that they will get more technical questions at the top of the funnel, more and more candidates are using AI to try and second-guess that. And so I'll get perfect answers to questions that should take a bit for someone to go, "Well, hang on, actually let me think about what my answer to this is," right?

The questions have to have more weight to them. They have to be a bit more open-ended to elicit more technical depth in the answers you're going to give me, and not be perfect. We both have to be more prepared, but more willing to accept flaws, because that's the weird dichotomy that's happening at the moment.

Q: Yeah, that's an interesting tension. In the writing world, I've heard of people intentionally putting flaws or typos in to be like, "I'm a person." I wonder if part of your prep is putting the questions into ChatGPT and being like, "This is a little too close to that answer."

A: Yeah! Literally that, or taking the transcript afterwards and just putting it in and saying, "Is this a perfect answer to this question?" and being like, "Yes"—because perfection is now a mark of fraud. That's an extreme word for it, but that's how you have to look at it.

So AI is helping you root out the perfection in a candidate, when what I think is more interesting are the flaws. Because again, what are we looking for here? We're looking for someone who wants to invest three to five years of their life in a company and get something out of it, right? That should be an open discussion. The idea of a job for life is sadly moving away from us in many industries.

So I very openly say to candidates: "You are obviously going to look at this as a place to get three to five good years of experience. What do you want from that experience that will set you up for that next stage of your career, based on what you've got now?" That shouldn't be a perfect answer. There should be doubting, but that also should be informed by the experiences they've had before. And that can be anything—"I want to be better at Databricks," or "I want to go into technical leadership". That is an example of a question that will tell me a lot more about a person than a binary "is this right or wrong" answer.

Q: That's the old "Where do you see yourself in five years?"

Q: "Celebrating the fifth anniversary of you asking me this question."

A: Exactly. There are better ways of asking that question. You can be really sensitive to the idea that a job is a necessity, but a career is a luxury. Let's admit that you'd love a job that paid you and let's establish that as the right answer. What are the nice-to-haves on top of that that keep you there?

And if you can have that information even before the call—if I could understand that about someone beforehand—the awkwardness is removed. I can go, "Hey, looks like you want to do this. Tell me more about that. Where does that come from? Can we offer that here? Is this a good fit for you"?

Q: Can you understand that sort of information from a resume?

A: No, because I don't think culturally we've embraced resumes as being the way to tell that story.

Someone at some point in the last 10 years said, "Fit it all onto one side of an A4." We still look at resumes as physical bits of paper that you would print off and read; we're still conditioned to how much fits in that very specific size. I have to remind myself sometimes that I think about these things because it's my job. My friends who don't work in this still think a resume is something you put on an A4 piece of paper with your dates of employment and a couple of things you did, and that's enough.

And actually, it's not enough anymore. That's table stakes. What I want to see is more depth and more of you as a person, because that's who we're hiring.

As I was saying earlier, I'm not expecting people to write code anymore. I'm expecting them to be active participants in the discussion of how something gets built. If you're an engineer, you have to be so much more product-minded than you ever had to before. If you're a product manager, you have to be so much more engineering-focused and closer to the code than before. Everyone's having to do 40% more than they did before because AI is automating a lot, but also giving you so much more information at the top.

As a result, a resume can't fully explain that shift in how people have worked in the last few years because we haven't really agreed on the syntax for it.

Q: The resume ain't quite what it used to be. All that information you put on a page, copied over to LinkedIn, copied over to application forms... it's not quite as useful as a marker of how good somebody is. Besides talking to the

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →