Research Staff, Data Science

Deepgram•Remote, CA

2d•$150,000 - $220,000

About The Position

Deepgram is seeking seasoned Data Scientists with demonstrated experience solving hard data problems while exploring research frontiers to join our Research Staff. Conversational audio presents incredibly rich scientific, engineering, and infrastructure challenges that are orders of magnitude harder than working with text. As a Member of the Research Staff, you will help us to build an industrial “data factory” that will be used to power the next generation of Voice AI systems - unlocking the creation of models that go beyond basic transcription and comprehension, capturing nuanced meanings in complex conversations, adapting robustly to diverse speech patterns, and generating empathic responses with human-like, contextualized speech. You will collaborate closely with our product, engineering, and data teams to build and deploy models in the most scalable voice API on the planet. We look forward to you bringing your expertise, sharing insights from your latest experiments, and collaborating with us to push the boundaries of AI and voice technology.

Requirements

Experience building data processing pipelines from a blank page and owning the entire data stack including data acquisition, characterization, cleaning, serving and transformation
Experience and expertise applying statistical methods and deep learning models to understand complex data
Strong communication skills and the ability to translate complex concepts in simple terms, depending on the target audience
Strong software engineering skills with particular emphasis on developing clean, modular code in Python and working with Pytorch

Nice To Haves

Background in Physics, Mechanical Engineering or Language Processing
Experience building models
Speech and audio experience

Responsibilities

Drive high performance data acquisition, preparation and synthesis pipelines to generate data for the next generation of speech and language AI foundation models
Develop advanced characterizations of complex conversational audio utilizing a diverse toolkit of signals processing techniques and deep learning models
Collaborate with DataOps and Engineering to create automated systems which scale the ability of human annotators to label high value data and provide critical feedback on model outputs
Build advanced benchmarking methodologies and curated datasets for evaluating conversational voice systems
Document and present results of data experiments and analysis for internal and external audiences