The role asks for someone who can build and run LLM agent workflows that turn spoken nursing notes into structured documentation, and who treats evaluation as a core part of the job. A strong candidate has shipped ML or NLP models to production, designs benchmarks and regression tests from real usage data, and can debug failures across app, backend and models. Solid software engineering habits and sound judgment with sensitive health data matter more than knowing the specific stack.
How it fits the AI era
Much of the drafting work moves to LLM agents, so the human work shifts toward defining context, tools and prompts, and judging when output is safe enough to ship. Careful evaluation, error analysis and privacy decisions stay human and become the main lever on quality. The role also weighs when smaller or self hosted models beat rented frontier ones on speed, cost or privacy.
How it builds AI experience
The job offers hands on work with LLM agents, structured outputs and context design in a live product used by many care workers, plus building benchmarks from production traces. It also gives practice training or adapting models for tasks like personal data detection and making latency, cost and privacy tradeoffs. Someone building toward this could start by shipping an agent workflow with a real evaluation suite and studying how failures are traced and fixed.
Worth showing
- Shipped NLP models to production
- Built LLM evaluation suites
- Agent and tool use
- Error analysis and regression tests
- Tested production code
This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.
A voice app that turns what a nurse says out loud during care into structured documentation, so less of a shift is spent typing it up afterward.
#J-18808-Ljbffr