AI workflow / full-stack implementation engineer

Compensation: No equity

We are a stealth healthcare AI startup building the operations layer for private medical clinic groups.
Medical clinics are drowning in repetitive administrative work: phone calls, scheduling, referrals, billing questions, prior authorization, follow-up, documentation, and handoffs across fragmented systems. These workflows are painful for staff, frustrating for patients, and expensive for clinic owners.
We are starting with a signed pilot customer: a multi-location private medical clinic group. Our first workflows are phone operations and prior authorization. The goal is not to build a generic chatbot. The goal is to build reliable workflow automation that works in the real world, with human review, auditability, escalation, and careful privacy/security controls.
We are early, moving quickly, and looking for people who are excited by messy operational problems. This is a chance to help build a vertical AI company from the ground up.

About the role:
We are hiring an AI workflow / full-stack implementation engineer to help turn messy clinic operations into working tools, evals, structured workflows, and implementation support.
This role may be structured as a paid internship, co-op, part-time role, or new grad contractor. It is a strong fit for an exceptional student, new grad, or early-career engineer from Stanford, UC Berkeley, Duke, UNC Chapel Hill, or NC State.
This is not a non-technical intern role, ops-only fellowship, or research role. The person needs to be a technical builder first, with enough product and operations judgment to work close to clinic workflows.
You do not need to own production architecture, but you do need to be able to write code independently and build useful systems under direction from the founding engineer.

What you will do:

• Map call types, handoffs, edge cases, staff rules, and escalation paths.

• Convert messy clinic processes into structured technical specs.

• Build eval datasets for phone agents using mock or approved de-identified data.

• Test agent outputs against clinic policies and workflow rules.

• Build internal tools for reviewing calls, workflows, escalations, and prior authorization status.

• Build lightweight dashboards, QA tools, or admin interfaces.

• Write data extraction, classification, and validation scripts.

• Support prior authorization workflow mapping and manual confirmation procedures.

• Help create workflow checklists, QA rubrics, and implementation docs.

• Translate operational details into clear documentation that engineers can build from.