Hire AI Software Engineer Intern — the employer’s brief

Insist on defining what "working" means before building, and on enough working inside someone else’s code style without arguing about it to work unsupervised on small tasks. Breaking a problem into shippable pieces without being told how is the third thing worth testing in the interview. Tool familiarity — Python, OpenAI or Anthropic APIs, scikit-learn or PyTorch — is a bonus rather than a filter: most of it is a week of learning for someone with the underlying skill.

Ship one prototype with a measured success rate rather than a demo. That is sized for eight to twelve weeks of supervised work by someone with the fundamentals and no production experience. A second, smaller piece — merge one reviewed change and see it run in production — usually fits alongside it. Anything requiring independent production judgement should stay with the reviewer.

Start from ₹20,000–₹49,500 a month, then adjust for city and duration: metros at the top, tier-2 typically 25–40% lower, and six-month commitments above six-week ones. Publish the number in the listing — "as per industry standards" is read as low or undecided, and it costs you applications from exactly the candidates who had another option.

Ask "How would you know this AI feature is working in production?" — you are listening for evaluation thinking — the difference between an engineer and a demo builder. Then follow the example they give rather than moving on to your next question. Score every candidate on the same set so the shortlist stays comparable.

One finished, reviewed piece of work that someone on the team would otherwise have done — not a side project nobody adopts. The deliverables above are sized for eight to twelve weeks of supervised work by a student with the fundamentals but no production experience. If they can demo it and the team keeps using it after they leave, the hire paid for itself.

The "AI" qualifier needs a measurable success definition before any build. Demos are easy in this space and production is not, so agree what "working" means in week one. In screening terms, that means adding one specific check: defining what "working" means before building.

The "Software" qualifier sets an expectation of code output. Have a repository and a first ticket ready, or you will lose the candidate’s momentum in the two weeks it takes to arrange access. In screening terms, that means adding one specific check: working inside someone else’s code style without arguing about it.

No. The posting assistant asks a few short questions — the role, the work, the duration, the stipend — and drafts the description, the title and the skill tags for you. You review and edit everything before it publishes, and you can paste in your own description if you already have one.

Yes. One listing is free and goes live after a quick company verification, usually inside two working days. Paid plans start at ₹499 for five postings a month, publish instantly with no review wait, and unlock every applicant's résumé and contact details. Both routes reach the same candidate pool.