Remote job
Research Engineer, Speech
Job details
About this role
Role overview This full-time research engineering role develops speech and audio models for contextual conversation, including interactions with pauses, overlap, interruptions, and changes of direction. You will work with model and inference engineers to make these behaviors effective in continuous, on-device use. The role is listed as based in San Francisco or remote.
Responsibilities - Design and train models for speech generation, audio representation, or conversational behavior, drawing on your area of expertise. - Build audio data workflows for cleaning, segmentation, alignment, and quality assessment. - Study how dialogue history affects prosody, delivery, and consistency between turns. - Develop turn-taking, backchannel, and interruption behaviors, including recovery when a speaker continues or changes a request. - Combine listening studies with objective measures of intelligibility, voice quality, timing, and interaction failures. - Improve streaming generation with systems engineers, evaluating responsiveness and output quality on target devices.
Requirements - Strong Python skills and hands-on experience with a modern deep-learning framework. - Depth in at least one speech or audio modeling area, such as synthesis, recognition, audio representations, or speech-to-speech systems. - Experience preparing audio data and designing experiments to investigate model behavior. - Ability to connect model training with inference, evaluation, and an interactive prototype. - Attentive listening and sound technical judgment about qualities conventional benchmarks may miss.
Nice to have Experience with streaming architectures, neural audio codecs, waveform generation, full-duplex systems, multilingual speech, expressive synthesis, human preference studies, or low-latency deployment.