Research Staff, Voice AI Foundations - Remote

Deepgram operates the leading platform powering the emerging voice AI economy, providing real-time speech-to-text, text-to-speech, and voice agent APIs used by over 200,000 developers and 1,300+ organizations worldwide. The company has processed more than 50,000 years of audio and transcribed over 1 trillion words, building foundation models deployed through cloud APIs and on-premises software.

Deepgram operates with an AI-first mindset where all team members actively use and experiment with advanced AI tools as part of their everyday work. The company moves at the pace of AI development, meaning your day-to-day responsibilities will evolve rapidly alongside technological change. This environment rewards those who embrace experimentation, adapt quickly, and continuously learn new capabilities rather than those seeking predictable, prescriptive work.

What you'll do

• Pioneer the development of Latent Space Models to solve fundamental data, scale, and cost challenges in building robust, contextualized voice AI systems

• Build next-generation neural audio codecs that achieve extreme compression with high fidelity reconstruction across diverse, real-world audio corpora

• Design steerable generative models capable of synthesizing human speech diversity, from casual conversation to emotional expression to multi-speaker scenarios with environmental noise

• Develop embedding systems that factorize codec latent space into interpretable dimensions like speaker, content, style, and environment, enabling precise control and synthetic data generation at scale

• Create model architectures, training schemes, and inference algorithms optimized for hardware efficiency that can handle billion-hour datasets while supporting real-time inference for hundreds of millions of concurrent conversations

• Design controlled experiments that isolate the impact of architectural innovations and validate theoretical insights

What you'll bring

• Strong mathematical foundation in statistical learning theory, particularly self-supervised and multimodal learning approaches

• Deep expertise in foundation model architectures and understanding how to scale training across multiple modalities

• Proven ability to bridge theory and practice through novel mathematical formulations and efficient implementation

• Demonstrated experience building data pipelines that process and curate massive datasets while maintaining quality and diversity

• Track record of identifying the critical experiments that validate or disprove ideas within days rather than months

• Experience optimizing models for real-world deployment with knowledge of hardware constraints and efficiency techniques

• Comfort identifying unsolved problems as opportunities to pioneer entirely new approaches

What they offer

• Remote work across The Netherlands and Melbourne, Australia locations

• Full-time position with a research-focused team at a well-funded Series C company