Recruiting an AI engineer: from profile to practical evidence

UTANA / WORKFLOW DESIGN

Hiring an AI automation engineer means looking beyond a list of tools. This proposed workflow uses your recruiting software to organise candidate profiles and answers into an evidence pack your team can review before the first screening call.

Define what you need to see

Start with four criteria: Python, multi-agent workflows, deployed automation, and a working demo. For each, ask for a concrete example of what the candidate built and their own contribution. A keyword match is a reason to investigate, not proof of experience.

Invite your shortlist to a conversation

Your team shortlists candidates in your recruiting software and approves a short invitation to answer practical questions before a screening call. Candidates who agree to participate are introduced to your company’s automated recruiting assistant, which gathers their answers through the software’s available messaging tools.

Ask about work they have actually built

The assistant asks one question at a time and follows up when an answer needs detail:

• What have you built with Python, and which parts did you write yourself?
• Describe a multi-agent workflow you built. What did each agent do, and how did they work together?
• How did you deploy it, where does it run, and how do you monitor or recover from failures?
• Can you share a working demo and a GitHub repository, or walk us through an example you are allowed to share?

It asks for specifics rather than accepting a list of tools. For example, “I built agents” prompts a follow-up about their tasks, inputs, outputs, and the candidate’s contribution.

Organise the replies for your team

When answers provide relevant examples against your agreed criteria, the agent stars the candidate and adds them to an internal “Potential screening” list. Your team sees the original answers, a concise summary, project links, and unresolved questions, then decides whether to arrange a call.

When replies do not yet provide enough relevant detail, the assistant thanks the candidate and records the conversation without adding a star. Unclear answers are marked for clarification. These records remain available for human review; the acknowledgement is not an automatic rejection.

A simple closing message could be: “Thank you for sharing your experience. I’m the recruiting assistant, and I’ve recorded your responses for the team. They’ll contact you if they would like to arrange a screening call.”

Measure before expanding

Before the pilot, record the time spent reviewing profiles and gathering project evidence. Compare it with preparation and human review time after automation. Track incorrect flags, missed relevant candidates, and follow-up needed. Without this baseline, time saved and mistakes reduced are difficult to verify.

Start with this one workflow. Once it produces useful evidence and measurable improvements, add interview scheduling or other recruiting workflows and agents.