Why now
Cheap code made judgment the bottleneck
AI made code cheap to write, not cheap to own. Everything your team ships still has to be upgraded, debugged and paged on for years. Cheaper building means more to maintain, and building the wrong thing costs more.
NVIDIA CEO Jensen Huang described the shift on Jodi Shelton's A Bit Personal podcast. Asked to define smart, he set aside the usual answer, someone who solves technical problems, and pointed somewhere else:
“People who are able to see around corners are truly, truly smart. And their value is incredible. To be able to preempt problems before they show up, just because you feel the vibe.”
In operations, that vibe has a source: years of watching systems break. It's the engineer who spots the shared dependency that will take down a whole region before a drill proves it, who stops a rollout at the first odd metric, and who decides what not to build. Huang added that this person “might actually score horribly on the SAT.” Most interview loops are the SAT.
Cheap code doesn't make senior operations engineers expendable. It makes them the bottleneck, and it makes hiring the right one the most consequential call you'll make this year.
Most interview loops test what's getting cheaper
Coding rounds measure how well someone writes code, and that's exactly the part getting cheaper. A loop built so anyone can run it falls into the trap Netflix co-founder Reed Hastings warned about: “if you dummy-proof the process, you only get dummies to work there.” A checklist interview selects for people who are good at checklists.
What you need to know about a Staff ops candidate is harder to see. Did they operate the system they describe, or sit next to it? Can they find where it breaks before you point at it? Would their automation survive your scale? In other words: would you hand them the pager?
When it makes sense
- You're hiring someone who will know more about operations than anyone on your interview panel.
- It's your first or second Staff SRE, platform or database reliability engineer.
- No one on your panel has run a large fleet: sharded databases, MySQL or Vitess at scale, Postgres fleets, homegrown control planes.
- You've just had an incident your last senior hire was supposed to prevent.
How it works
Calibration call 45 min
Your systems, your scale and what “senior” has to mean on your team. It becomes an interview plan for your role.
The interview 45 min, Zoom
The candidate picks the system they've owned that best shows their operational work, and the conversation goes deep: architecture, failure modes, what they'd change, what they built themselves, and how they'd rebuild it. If there's a take-home or open-source work, it covers the choices they made in it.
Written assessment next business day
A recommendation calibrated to your scale, the evidence behind it, where they'd need support, and the questions the rest of your loop should ask. See a sample assessment →
Debrief 30 min
Walk through the assessment, pressure-test it against the rest of your loop, and get your questions answered.
What the interview listens for
Owned it, or sat next to it
Plenty of strong resumes describe systems the candidate worked near. The interview separates what they designed and built from what their team did.
Where it actually runs
What runs where, how a component lands on a host, the real numbers. “It's all in the cloud” isn't an answer.
Finds the weak spot unprompted
Senior engineers name the single point of failure and its blast radius before anyone points at it.
Runs the fleet, not just the diagram
A platform that has provisioned a handful of clusters in a fleet of a thousand is a design, not yet a track record.
Chooses for the scale, not for familiarity
Knowing an approach won't scale and choosing it anyway because it's comfortable is a signal, and so is knowing why the alternative wins.
Knows where humans still glue it together
Every automated system has a step a person still does by hand. Strong candidates know where theirs are and what they'd do about them.
Gets clearer the further you drill
The engineer you want at 3 a.m. explains a system more precisely as the questions get harder. If the project they picked can't carry the depth, the interview moves to a design problem without the safety net of their own history.
Want to run this round yourself? The full guide: five questions, what to listen for, and how to write it up →
Who runs it
Vitaliy Mogilevskiy
Principal · Staff Platform Engineer
At Slack, Vitaliy led more General Tech & Automation interview rounds than anyone on his team: 20+ Senior and Staff candidates from 2022 to 2026, judged against the bar for operating one of the largest Vitess fleets. He brings ten years of database platform automation at InVisionApp, Auth0 and Slack, and an Oracle DBA background before that. He built the Vitess upgrade control plane at Slack, now open source as hammerdown, and is a named inventor on a patent for cloud database provisioning.
How he works: principles at mogilevskiy.dev →Price
$2,500 for the first candidate on a role, including the calibration call. $2,000 for each additional candidate on the same role.
- Invoiced when you book. Rescheduling is free with 24 hours' notice.
- The written assessment arrives by the next business day after the interview.
- You make the hiring decision. The assessment is an independent technical input to your process.
Questions
Does this replace our interview loop?
No. It's one round: the one most teams can't staff themselves. Your team still owns coding, collaboration, culture and the final call.
Do you run coding interviews?
No. Keep your existing coding round or take-home. If there is one, the interview asks the candidate about the choices they made in it.
Which roles fit?
Staff and Senior SRE, platform, infrastructure, database reliability and operations-automation engineers. It's not for junior or mid-level roles, or for high-volume screening.
What if our stack isn't MySQL or Vitess?
The method is about operating any large system, not one product. If a role needs deep expertise Vitaliy doesn't have, you'll hear that on the calibration call, before you pay for a candidate.
What does the candidate hear about us?
You introduce Vitaliy as an independent interviewer you've engaged. HASHJOIN never recruits your candidates or your employees.
Is the interview recorded?
No. The assessment covers job-related technical and operational evidence only. Resumes, notes and drafts are deleted 90 days after delivery.
Can you work through our vendor process?
Yes. HASHJOIN contracts business-to-business, and an NDA before the calibration call is fine.
Hiring a senior ops engineer?
Tell us about the role and the timeline. You'll get a straight answer about whether the Pager Test fits, usually within one business day.