Research Staff, Voice AI Foundations [Remote]
Salary not listed • Remote • Australia- Melbourne • Full-time
Posted 20d ago
Remote Work Policy
Fully remote
Employment Type
Full-time
Categories
AI Research Engineer
About the job
Deepgram is a leading platform for Voice AI, providing real-time APIs for speech-to-text and text-to-speech, enabling developers to build production-grade voice agents at scale. The company is at the forefront of voice technology, processing vast amounts of audio and transcribing trillions of words. Deepgram fosters an AI-first mindset, expecting all team members to actively use, experiment with, and integrate advanced AI tools into their daily work, embracing rapid change and continuous learning. This role offers the opportunity to pioneer new paradigms in audio AI, addressing fundamental challenges in data scarcity, scale, and cost to make voice interaction universally accessible.
Responsibilities
- Develop next-generation neural audio codecs for extreme low bit-rate compression and high-fidelity reconstruction.
- Pioneer steerable generative models for synthesizing diverse human speech from latent representations.
- Develop embedding systems to factorize codec latent space into interpretable dimensions for precise control.
- Leverage latent recombination to generate synthetic audio data at scale for joint model and data scaling.
- Train multimodal speech-to-speech systems for understanding diverse human inputs and producing empathic responses.
- Design model architectures, training schemes, and inference algorithms optimized for cost-efficient training and real-time inference.
Requirements
- Strong mathematical foundation in statistical learning theory, especially self-supervised and multimodal learning.
- Deep expertise in foundation model architectures and scaling training across modalities.
- Proven ability to bridge theory and practice, deriving novel mathematical formulations and implementing them efficiently.
- Demonstrated ability to build data pipelines for processing and curating massive, high-quality datasets.
- Track record of designing controlled experiments to validate theoretical insights and architectural innovations.
- Experience optimizing models for real-world deployment, considering hardware constraints and efficiency.