Precision
How many of the people we send are worth interviewing. Three to five candidates with written assessments, rather than a long list that puts the screening work back on you.
precision = TP / (TP + FP)
Precision & Recall recruits AI and data engineers in Australia. Both founders have built and run these systems, which is why we assess candidates ourselves rather than passing CVs along.
Technical hiring is hard to get right from either side. A hiring manager reads a stack of CVs with little in them to separate one from the next. An engineer gets an approach and cannot tell from it whether the job is worth an hour of their evening.
Both come down to the same thing. Telling apart the engineer who has run a system in production from the one who has read about it is a technical judgement, and it is a much easier one to make if you have done the work yourself.
So we kept the part of recruitment that needs judgement and did away with the part that does not. We work a small number of searches at a time, and a founder runs every technical conversation.
One founder is a Chief Data & AI Officer with a research background in deep learning and a career through enterprise transformation and GM roles in AI. The other has spent twelve years building production data platforms.
Between them they have specified these roles, interviewed for them, and lived with the consequences of getting them wrong.
Precision and recall are the two measures used to judge a machine learning model, and they pull against each other. You can send everybody and be sure the right person is in there somewhere. You can send almost nobody and be sure everyone you send is good. Neither is useful on its own.
How many of the people we send are worth interviewing. Three to five candidates with written assessments, rather than a long list that puts the screening work back on you.
precision = TP / (TP + FP)
How many of the people who could do the job we actually reached. Most of them are employed and not applying anywhere, so this is the part that takes the work.
recall = TP / (TP + FN)
Whether you are hiring or thinking about your own next role, we will tell you straight whether we can help.
Get in touch