Hiring for Engineering at Savvy
I recently posted an engineering role and had hundreds of applications in under 24 hours. When I started reviewing them, more than half had pulled verbatim wording from the JD into the resume, and a quarter looked identical. Same format, same font, same bullet structure, just different names at the top. My first reaction was to literally throw my face in my hands, and my first thought was that every signal I used to rely on was no longer reliable.
The conversation around AI and hiring is usually about volume, but volume is a problem with an easy solution. The harder problem is that the signals themselves moved. A polished resume, a clean take-home, fluency with the right stack, none of it tells you what it used to, because AI flattened all of it. So the question I've had to answer is what actually predicts a great engineering hire now, especially at a company like Savvy where we're building an AI-native system and every employee, even non-engineers, works with AI daily. Here's where I've landed (and I’m sure leadership will be happy to see how well this lines up with our company values:
What I look for now
The first thing I listen for in a screen is how someone relates to the change happening in the industry.
The engineers who do well here are the ones who saw AI shifting how software gets built and moved toward it, because they wanted to be part of building the next big thing rather than waiting to see how it plays out. That's Invent the Future in practice. Healthy skepticism is also necessary, but the version of skepticism we want comes from experimenting, not from the sidelines. Our engineers aren't afraid to make mistakes. The expectation is that you let go of your ego, learn from what went wrong, and come back better, which is what Compound Daily actually means day to day. The candidates I get most excited about can tell me how their own workflow has changed in the last six months, what they tried, what failed, and what they do differently now.
The second thing is how someone works with the rest of the company.
The engineers who thrive at Savvy aren't the ones who sit behind a keyboard for ten hours a day and wait for requirements to arrive. They're in the room with product, with the business teams, sometimes with advisors themselves, digging into what we're building and why it matters before they write a line of code (or, now, prompting the agent to write it for them). That's Win Together and Seek Truth working side by side: leave your ego at the door, debate with conviction, and question relentlessly until you understand the problem well enough to solve the right version of it. In a screen, this shows up in small ways. Candidates who can tell me what their last company's customers actually cared about tend to be builders in this mold. Candidates who can only describe their slice of the codebase usually aren't.
The last thing I look for is whether someone can move on their own.
At 35 engineers building a platform that manages billions in client assets, there's no layer of people whose job is to hand you priorities, check your work, and clear your path. The engineers who thrive here see a problem, decide it's theirs, and drive it end to end with urgency, because they act like owners rather than employees waiting for direction. That's Own the Outcome, and it's one of the values I probe hardest for in a screen and the team continues to test throughout the process, because autonomy isn't something you can coach into someone after they start, and it isn't something AI can fake for you in a hands-on interview. But autonomy without awareness is just recklessness with confidence, so the best candidates carry conviction in their work while still thinking about who their decision affects, who should know before it ships, and who to pull in when they're at the edge of their judgment. If the story of a decision they made alone includes the moment they looped someone in, that's an owner. If the story is only about how they didn't need anyone, that's a flag.
Coming full circle
Notice that none of these signals show up in a resume. You can't paste them from a JD or generate them in one shot, which is exactly why they're the ones left standing. And that's the irony of where hiring has landed. I use AI every day, and so does everyone at Savvy. It's in how I source, how I prep for screens, how our engineers build, how the whole company runs. The tools were supposed to make evaluation more scientific, and for a while it felt like they would, but the more AI leveled the playing field on resumes, take-homes, and polish, the less those things could tell us. We still use the numbers and metrics, they're part of every process we run, but they've gone back to being inputs rather than answers. What we've really come back to is the oldest tool in recruiting: sitting across from a person, asking the right questions, and trusting our own judgment about who they are and how they think. AI changed almost everything about how we hire. What it couldn't change is that the final call was always human, and in 2026, it's more human than it's been in years.
