Prompt Engineer

Prompt engineers design and optimize the instructions that go into LLMs to get reliable, high-quality outputs. It's part writing, part experimentation, part light engineering — they build prompt templates, evaluate model behavior at scale, and figure out how to make an unpredictable model behave predictably for real users.

Writing a new prompt template, running it across an eval set, comparing GPT-5 vs Gemini outputs, debugging why the model started ignoring the system message, and shipping a tweak that bumped accuracy 4%.