A complete application in a minute
Tailored resume and cover letter, ready to send.
present of AI in San Francisco seeks a mid-level Customer & Solutions specialist who sits between customers, research, and data production. You turn customer questions into fast experiments, read results, and adapt audio data collection or filtering based on model outputs.
You code well, understand model evaluation, and can trace a model outcome back to recording setup, annotation, or quality checks. The role offers hands-on practice with ablations, eval harnesses, and linking data quality to
Customer & Solutions · San Francisco · Full-time · Mid level
This role sits between customers, research, and data production. It asks for someone who can turn a customer question into a fast experiment, read the results, and change how audio data is collected or filtered based on what the model shows. A strong candidate codes well, understands model evaluation, and can trace a model outcome back to recording setup, annotation, or quality checks.
The work is built around making AI models measurably better through data, so AI is the product and the measuring stick. Much of the loop is automated, but judgment about which experiment matters, why a model regressed, and what to recommend to a customer stays human. Speed of iteration and clear communication carry more weight than routine coding.
A person here would gain hands on practice running ablations, building eval harnesses, and linking data quality to model behavior for speech and audio models. That experience maps directly to the company focus on high fidelity conversational audio datasets for voice foundation models. Building it means getting comfortable with training, fine tuning, and evaluation workflows plus the operational side of how audio gets recorded and checked.
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.
High-fidelity conversational audio dataset generation and speech research platform for foundation voice models, per the company.