Assignment: Develop a full-stack feature that simulates a co-pilot experience for Prior Authorization.

Our tech stack: Next.js, Tailwind, Python (FastAPI/Django), AWS, Postgres


A Utilization Management Nurse (UM Nurse) working for a Payor conducts nearly 100 case reviews per week. Co:Helm’s co-pilot enables UM Nurses to complete reviews much faster by offloading the administrative work in this process to a LLM pipeline. In short, the LLM pipeline takes a medical record PDF and extracts evidence from the document using a set of instructions to guide what needs to be found. The output from the LLM pipeline is then displayed to the UM Nurse on a web-based Dashboard for review. The outcome of a review is typically “Met” or “Not Met” based on the evidence found in the medical record.

You will be assessed on:

Submission instructions

We expect this task to take between 3 hours and a whole day, depending on your proficiency and the level of polish you apply. Submissions without a README or instructions on how to run the project will be rejected. If you would like to keep your GitHub repository private, please invite zaarheed as a collaborator. Short Loom video demos highly encouraged!

Your task

Fork and clone the starter repository (cohelm/product-engineer-starter) to complete the three exercises below, and choose one bonus feature to implement. Your submission will be screened on the following criteria before being reviewed: