Winter Internships
The AI Agent Builder Intern builds AI employees for real clients: defining one role end to end, training the agent on the client's playbook, setting what it may do alone and testing it until it is reliable. You work with the engineers who own each agent after launch.
Turn a client's SOPs and documents into a private knowledge layer that cites its sources.
Design an agent's tools, boundaries and escalation rules with the team.
Build evaluation sets from real cases and run the agent against them until it passes.
Log, review and tune agent behavior after launch.
You have built an agent, a bot or a retrieval system and can explain where it failed.
You think in edge cases and like writing them down.
You can read a messy SOP and turn it into rules a system can follow.
Email hello@automagic.agency with the role in the subject line, a few lines on what you have built or automated, and links to anything we can look at. Applications are reviewed as they arrive. Compensation and schedule are discussed in the first conversation.
Reads a client's playbook, designs the agent's tools and boundaries, builds test cases from real examples, runs the agent against them, and tunes it until the results hold. After launch, reviews logs and improves it.
Whatever fits the client's build. Interns are expected to learn tools quickly rather than arrive knowing one stack, and the team pairs on the first builds.
Yes. Interns work on live builds with review from the engineers who own them. Client names are kept confidential, as they are across the site.
automagic.agency
Begin with a scoping session
