SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
nyra health is seeking a Researcher in Post Training to develop methods that make speech models more accurate, controllable, robust, and aligned with real-world clinical needs. This is a research role with strong engineering ownership, where you will take ideas from hypothesis through experimentation, evaluation, and release.
You will shape post-training methods including supervised fine-tuning, preference optimization, distillation, feedback-driven learning, and reinforcement learning. You'll create high-quality post-training datasets through curation, annotation, synthetic data generation, and model-assisted improvement. Your work will improve model behavior across instruction following, verbatim transcription, uncertainty handling, long-form consistency, multilingual performance, and hallucination resistance.
You will build benchmarks and failure taxonomies to reveal genuine model improvements, implement and scale training and evaluation pipelines with reproducibility focus, and adapt foundation models to specific speech and clinical use cases. You'll contribute to publications, model releases, datasets, and technical reports.
nyra health operates in neurorehabilitation, providing software that supports clinics, therapists, and patients. nyra labs is the research arm, turning difficult clinical problems into open models, datasets, benchmarks, and research for the wider community. You'll have access to a uniquely large, professionally labeled neurological speech dataset with millions of recordings from real clinical settings.
Success requires post-training expertise (fine-tuning, preference optimization, RLHF, distillation, alignment), strong PyTorch and modern model-training workflows, experience designing evaluations and diagnosing complex model behavior, ML engineering ability with clean production-quality Python and distributed training debugging, and a record of publications, open-source work, or substantial independent research. MSc, PhD, or equivalent practical experience in machine learning, speech processing, NLP, or related field is expected. You should be scientifically honest, behavior-focused, impact-oriented, self-directed, and collaborative across research, engineering, product, and clinical teams.
