You are the IC who got promoted. You can read code, ship reviews, and ship interviews, but not all in the same week. We screen, interview, score, and recommend, against a rubric that understands what good actually looks like. You meet three engineers. You pick one.

"I have 187 unread applications and a sprint review at 11. One of them is probably brilliant. I will not find them today."

"Doing screening calls at 10pm because we keep losing the good ones to faster companies. This is unsustainable."

"Recruiter sent six "great" CVs. None can describe a system they've built. We are not hiring CV-writers."

Conversational probe on a realistic system. We score against a published rubric: comfort with eventual consistency, capacity reasoning, failure modes. Not jargon density.

A small, scoped take-home or live exercise (configurable). We score for readability, edge-case handling, and how the candidate reasons about the trade-offs they made.

Did they fight for reliability when no-one asked? Did they own a thing past v1? We probe this in the interview and grade specific examples.

How quickly do they reach for the boring, working answer? We listen for "I would just" and what follows.

A short discussion of a time they helped someone else grow. Optional at IC1/2, default-on at IC3+.

Hard filters from your post: language, stack, region, level. We do these first, before the model spends any time on the candidate.

247

Inbound applicants screened

94%

Top-3 conversion to interview

31h

Median time from post to shortlist

4.6h

Median time you spend hiring

"Posted Monday morning. Wednesday I interviewed three engineers. Thursday I had an offer out. The candidate started the next month. I have not done that in six years of hiring."

"My recruiter and I were doing twelve screening calls a week between us. Now I do three real conversations. Better engineers, less Sunday work, lower spend."

"The rubric is the part I was most sceptical about. Then I read three of them. It would take me an hour to write notes that thorough."

"We compared two hires we made through Picked against two we made through our usual loop. Both Picked hires are still here. One of the others is not."