Building and Evaluating Clinical Speech AI with NVIDIA, Nebius and J.P. Morgan
Explore how NVIDIA’s open models can power ambient-care and clinical voice-agent workflows.
This session, supported by J.P. Morgan and NVIDIA, offers a practical walkthrough of deploying Nemotron Speech on Nebius, adapting it to medical vocabulary and evaluating its accuracy and safety for clinical use. Through a live demo and Q&A with engineers, we’ll discuss how to integrate these models into your applications and what to consider when planning for production, including data privacy.
What you’ll learn
- How to deploy Nemotron Speech on Nebius
- How to adapt speech recognition to medical terminology
- How to evaluate accuracy and safety in clinical workflows
- What to consider for integration, data privacy and production readiness
Who should attend
Developers, product managers, clinical informaticists, research scientists and ML/NLP engineers building clinical AI applications, along with founders and technical executives at AI-native digital-health companies.
Familiarity with AI models is helpful; no prior experience with Nemotron is required.
Our speakers
Cedric Steenbeke
Register to get the recording
Key takeaways
Clinical speech recognition must capture medical terms, drug names and dosages accurately while handling accents, background noise and brief responses.
The webinar explains why these requirements matter when evaluating models for clinical conversations, and how latency and speaker identification affect their usefulness in healthcare applications.