How to Write a Job Description That Attracts Top AI Talent
The market for AI and ML engineers is brutally competitive. The best candidates have inbound offers weekly and ignore most job postings on sight. If your job description looks like every other one — long lists of bullet-pointed duties, vague 'exciting opportunity' copy, and a wall of required skills — you've already lost them. Here's what to do instead.
Why Most JDs Fail to Attract AI/ML Talent
Top AI engineers are problem-solvers first. They want to know what hard problem they'll be working on — not a list of meetings they'll attend. Most JDs read like HR compliance documents: generic responsibilities, inflated requirements ('5+ years of PyTorch' for a two-year-old library), and zero signal about technical depth. The result? Strong candidates scroll past, and you end up with a pipeline of applicants who applied to 50 listings the same afternoon.
The Sections That Actually Matter
Three things move the needle for AI candidates: the problem to solve, the tech stack, and the growth opportunity. Skip the corporate boilerplate and be specific. What data are they working with? At what scale? What does the model pipeline look like today, and where does it need to go? What infrastructure decisions will this person own? AI engineers want to know they'll be doing real work, not maintaining a YAML config file. On growth: be honest about what the role can become. A senior engineer joining a Series A has real leverage to shape the technical direction. Name that explicitly.
Red Flags That Make Top Candidates Skip
Experienced AI candidates are pattern-matching for warning signs. Watch out for: requirements lists longer than 15 bullet points (signals that you don't know what you actually need); phrases like 'wear many hats' without specifics (signals disorganization); vague compensation ranges or none at all (signals low trust); no mention of compute budget or tooling (signals the company hasn't thought seriously about AI infrastructure); and overly polished marketing language where technical substance should be. If your JD reads like it was written by a committee of non-engineers, it probably was — and candidates will notice.
A Template Structure That Works
Use this structure for your next AI/ML job description: Opening hook — one sentence on the problem your team is solving and why it's technically interesting. Role purpose — what this person will own and why it matters to the business. What you'll work on — three to five concrete examples of projects or challenges (be specific). Stack and tools — the actual tech: frameworks, cloud infra, data platforms, experiment tracking. What good looks like — what a strong hire accomplishes in their first 90 days. Why join us — what makes this role worth leaving a competing offer for: team caliber, technical autonomy, compensation, mission. That's it. No need for an HR-speak preamble or a five-paragraph company history.
Writing a great job description takes 90 minutes of focused thinking. It's one of the highest-leverage things you can do before you start sourcing. Get it right, and qualified candidates self-select in. Get it wrong, and you'll spend weeks manually screening a mismatched funnel.