Speech Scientist
Zyoin Group
Job Description
Responsibilities:Define and execute the long-term technical vision and roadmap for speech science research and developmentLead the design and development of advanced speech recognition, synthesis, speaker verification, and natural language understanding systemsDrive innovation by staying at the forefront of academic research and translating state-of-the-art techniques into production-ready solutionsMentor and guide senior and junior scientists, fostering a culture of technical excellence and intellectual curiosityCollaborate cross-functionally with product, engineering, and design teams to align research efforts with business objectivesRepresent the organization at conferences, publish in top-tier venues, and contribute to the broader research communityEstablish best practices for experimentation, model evaluation, and data pipeline developmentInfluence organizational strategy by providing expert input on build-vs-buy decisions and technology partnerships
Qualifications:Ph.D. / Masters with a background in Speech Processing, Computational Linguistics, Electrical Engineering, Computer Science, or a closely related field3+ years of industry experience in speech and audio processing, with a proven record of shipping products at scaleDeep expertise in end-to-end ASR, TTS, speaker diarization, speech enhancement, speech translation or related domainsProficiency in Python and deep learning frameworks (PyTorch, TensorFlow); experience with large-scale distributed trainingDemonstrated ability to translate ambiguous business problems into well-defined research agendasExcellent communication skills with the ability to influence technical and non-technical stakeholders at all levelsExperience leading or mentoring teams of scientists and engineer
s
Nice to Have:Experience with large language models and multimodal architecturesFamiliarity with on-device/edge deployment and model optimization techniquesTrack record of securing patents in speech or audio technologyStrong publication record in popular speech conferences / journa
ls