Research Staff – Voice AI Foundations (Remote)

About the role

Deepgram is looking for a Research Staff member to pioneer Latent Space Models (LSMs) that address data, scale, and cost challenges in voice AI. You will work on cutting‑edge audio research to enable universal, high‑quality voice interaction.

Key responsibilities

• Design and develop next‑generation neural audio codecs for extreme low‑bit‑rate compression and high‑fidelity reconstruction.
• Pioneer steerable generative models that synthesize diverse human speech, including emotional and multi‑speaker scenarios.
• Create embedding systems that factorize codec latent space into interpretable dimensions such as speaker, style, and environment.
• Leverage latent recombination to generate synthetic audio data at massive scale.
• Train multimodal speech‑to‑speech systems that understand any human and produce empathic, task‑oriented responses.
• Design model architectures, training schemes, and inference algorithms optimized for bare‑metal hardware to enable cost‑efficient training and real‑time inference.

Required profile

• Strong mathematical foundation in statistical learning theory, especially self‑supervised and multimodal learning.
• Deep expertise in foundation model architectures and scaling across modalities.
• Proven ability to bridge theory and practice, deriving novel formulations and implementing them efficiently.
• Experience building data pipelines that process massive, diverse audio datasets.
• Track record of designing controlled experiments that validate architectural innovations.
• Experience optimizing models for real‑world deployment, including hardware constraints.
• History of open‑source contributions or research publications advancing speech/language AI.

Required skills

• Statistical learning theory
• Self‑supervised learning
• Multimodal learning
• Foundation model architecture design
• Large‑scale data pipeline engineering
• Hardware‑aware model optimization