Latest / The Tech Career Podcast with Fexingo: Engineering Jobs, Interviews, and FAANG Career Strategy / How FAANG Engineers Interview Without Code
Transcript
- Lucas: So Meta ran a quiet experiment last year — a pilot program where senior engineer candidates could opt for a coding-free interview track. No LeetCode, no whiteboard, no coding challenge at all. Luna: Wait — zero coding? How do you assess someone's ability to build software without watching them write it? Lucas: Exactly the question. They replaced it with an extended system design session, an architecture walkthrough, and a deep dive into one of the candidate's past projects — like a mini case study. The idea is that for senior roles, you're hiring for decision-making and pattern recognition, not syntax recall. Luna: That's a pretty bold shift. Did it work? Or did they scrap it? Lucas: They actually reported a 20 percent reduction in false positives — candidates who aced the coding test but underperformed within six months. About 30 percent of senior engineer interviews in 2025 used that track. And now Google and Amazon are looking at similar models. Luna: You know, that stat about false positives — that's huge. And it's exactly the kind of thing that makes this podcast useful. If today's conversation gave you something to think about, the way we keep this show ad-free is through listener support. You can buy me a coffee dot com slash fexingo. That link helps us keep digging into trends like this without any sponsorships. Lucas: Yeah, and it really does make a difference. So back to the coding-free interviews — what does this mean for someone who's been grinding LeetCode for months? Luna: I think a lot of engineers are going to feel both relieved and anxious. Relieved because the interview prep becomes more about real engineering, but anxious because system design and architecture walkthroughs are harder to practice in isolation. Lucas: Right. LeetCode has a clear right answer. System design is open-ended — there's no single correct solution. It's more about trade-offs: why did you choose this database over that one? How do you handle failure at scale? Luna: And the past-project deep dive — that's even more personal. You have to be ready to talk about what you actually built, not just a contrived problem. Lucas: Exactly. Meta's pilot specifically asked candidates to walk through a project where they'd made a significant technical decision. The interviewer would probe — 'Why that API? What were the alternatives? How did you measure success?' — almost like a code review of your career. Luna: So it's more about storytelling and justifying past choices than solving puzzles. That's a different skill set entirely. Lucas: It is. And it might reduce bias too. Research has shown that whiteboard interviews disproportionately disadvantage women and underrepresented minorities. A 2024 study from the National Bureau of Economic Research found that removing coding tests increased offer rates for women by 15 percent at one large tech company. Luna: But does it just shift the bias? System design interviews can be more subjective. If the interviewer doesn't like your particular approach, they might mark you down unfairly. Lucas: That's the big concern. Meta addressed it by using a rubric — they scored candidates on four dimensions: problem decomposition, trade-off reasoning, communication clarity, and scalability awareness. And they trained interviewers specifically on this format. Luna: Still, I wonder if this is only for senior engineers. What about mid-level or junior roles? Those tend to rely more on coding assessments because you need proof that someone can write production-quality code. Lucas: You're right. Meta's pilot was exclusively for L6 and above — roughly staff engineer level. For juniors, they still use coding tests. But there's talk about extending a modified version to senior mid-level, L5, where the coding bar is lowered and system design is weighted more. Luna: What about Google? You mentioned they're looking at similar models. What's their approach? Lucas: Google is experimenting with what they call 'Googleyness and engineering judgment' interviews — essentially a behavioral round focused on technical decision-making. They ask things like: 'Tell me about a time you chose one technology over another and it backfired.' It's less structured than Meta's approach but similar in spirit. Luna: Amazon has always had their 'bar raiser' who looks for leadership principles, but they still require coding. Are they changing that too? Lucas: Amazon is piloting a 'portfolio interview' where you submit a significant contribution — a design doc, a pull request, a system diagram — and then discuss it for an hour. No live coding. But it's only for L7 principal engineers so far. Luna: So the takeaway for someone listening who's aiming for a senior role at a FAANG: should they stop studying algorithms altogether? Lucas: I wouldn't say stop. Companies still use coding assessments for screening — especially Google, which hasn't fully committed. But the trend is clear: if you're senior, invest more time in system design and articulating your past work. Have a few projects ready that you can talk about in depth. Luna: How do you even prepare for that? Practicing system design with a friend or a whiteboard is one thing, but the deep dive on your own project — that's harder to simulate. Lucas: I'd recommend doing a written retrospective on two or three of your biggest projects. Write down the key decisions, the alternatives, and the outcomes. Then practice explaining it out loud — record yourself. You want to sound confident, not rehearsed. Luna: And for the system design part, any resources you'd point to? Lucas: The classic is 'Designing Data-Intensive Applications' by Martin Kleppmann. Also, YouTube channels like 'System Design Interview' by Gaurav Sen. But the best practice is to actually design a system with someone who pushes back on your choices. Luna: One thing I think is underrated: learning to say 'I don't know' or 'I'd need to benchmark that.' In a coding interview, hesitation can be penalized. In a design interview, acknowledging uncertainty and showing how you'd resolve it is actually a strength. Lucas: That's a great point. The rubric Meta used explicitly rewarded candidates who could identify risks and propose follow-up experiments. It's more like a real engineering conversation. Luna: So is this the death of the coding interview? Or just a new lane? Lucas: I think it's a new lane — for now. But if Meta's results hold up, we could see coding tests become optional for more levels. The real question is whether other companies can replicate the training and calibration needed to make subjective interviews fair. Luna: And whether candidates will trust that process. There's a lot of skepticism that removing the 'objective' coding test opens the door to bias. Lucas: Right. But the data from Meta shows that the no-coding track actually improved diversity metrics — more women and underrepresented engineers passed. So it's not necessarily a step backward. Luna: I guess we'll see if this spreads beyond FAANG. Smaller companies might adopt it to compete for senior talent who don't want to study LeetCode again. Lucas: That's already happening. Some well-funded startups are advertising 'no coding interviews' as a perk. It's a hiring differentiator. Luna: Alright, so if I'm a senior engineer looking to switch jobs in 2026, what's the one concrete thing I should do this week? Lucas: Pick one project from the last two years — ideally one where you had to make a tough trade-off — and write a one-page design summary. Then schedule a mock interview with a friend or mentor where you walk through it for 45 minutes. That's your new interview prep. Luna: And if you're a junior engineer worried about this trend? Lucas: Don't panic. Coding tests aren't going away for you anytime soon. But start building that habit of reflecting on your work — keep a technical journal. When you get to senior level, you'll have a rich portfolio of stories ready to go. Luna: That's solid advice. I'm curious to see where this goes — especially if Google and Amazon roll out their pilots more broadly. Lucas: Same. It feels like the industry is finally asking: 'Are we testing the right thing?' And the answer might be no.