AI-Native Engineering Intern

At Marham, we generate and pressure-test ideas with AI, quickly validate the strongest ones against real patient and doctor signals, build the ones that hold, and ship them. You will write code that carries an idea through that loop, from a rough concept to working software that real users use within weeks.

Recruitment Process:
We do not read resumes. What you automate is how we evaluate you. Before any interview, we ask for one thing you have built.

The build. In 48 hours, take a real manual workflow (yours, a past employer’s, a society’s, a family business’s) and automate it end to end using AI agents. Examples: an agent that triages an inbox and drafts replies, a scheduled task that pulls data and sends a daily summary, a skill that turns messy inputs into a clean report, or a knowledge bot over a set of documents. It should actually run, not just be a diagram.

The constraint. AI agents and tools should do most of the work, both in building the automation and in running it. The point is to see how you think about workflows and direct the tools.

What you submit:

• A link to the working automation, or a screen recording of it running.
• A screen-recorded walkthrough of five minutes or less. Show the before (the manual process) and the after (the automated one). Then show how you set it up: the prompts, skills or connectors you used, where it broke, and how you fixed it.
• A one-page SOP that someone else could follow to use or maintain it.
• Three sentences on which Marham workflow you would automate first, and why.

We read every submission. We are looking for judgment in which workflow you picked, clarity in how you broke it down, and honesty about what still needs a human

Work Location: In person