Research Staff, Applied Voice AI

Role overview Improve voice AI systems by taking customer-facing model limitations from initial investigation through measurable production results. You will turn loosely defined problems into evaluation targets, select a practical approach, and ship improvements on a short cycle. Example areas include language coverage, speech synthesis pronunciation, speaker separation in noisy audio, and new voice capabilities.

Responsibilities - Identify where models fall short and define measurable improvement goals. - Design evaluations that show whether a change works and help explain why. - Curate or generate training data and fine-tune models as needed. - Explore suitable approaches, including new data or external models, to address customer problems. - Build working systems and release improvements for production use. - Use customer feedback and observed results to guide further research and iteration.

Requirements - Deep experience training, fine-tuning, and evaluating deep learning models. - Strong Python and PyTorch skills. - A track record of deploying machine learning systems in production. - Ability to own a problem through to a measurable outcome and learn unfamiliar methods as needed. - Sound judgment in prioritizing customer needs and deciding when a result is ready to ship. - Evidence of solving difficult problems through shipped projects, systems, or open source work; a specific degree is not required.

Nice to have - Experience with speech recognition, speech synthesis, or voice agents. - Distributed training experience. - Research publications, open source contributions, or public benchmarks.

Benefits and work setup The work environment values frequent experimentation with AI tools, adaptability, and continuous learning. Priorities and day-to-day work may change quickly, and the role emphasizes getting useful improvements to customers within weeks.