Research Staff, LLMs

📋 Description

• • Deepgram is at the forefront of the rapidly expanding trillion-dollar Voice AI economy, providing essential real-time APIs for speech-to-text (STT), text-to-speech (TTS), and the development of production-grade voice agents at scale. Our technology powers voice offerings for over 200,000 developers and more than 1,300 organizations, including industry leaders like Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. We offer unmatched accuracy, low latency, and cost efficiency through cloud APIs and self-hosted/on-premises software, built on our proprietary voice-native foundation models. With a recent Series C funding round led by top global investors and strategic partners, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words, solidifying our position as the world's foremost expert in voice technology.
• • At Deepgram, we embrace an AI-first operating rhythm. Proficiency and comfort with AI are not optional; they are fundamental to how we innovate, operate, and measure success. Every team member is expected to actively utilize and experiment with advanced AI tools, integrating them into their daily workflows and even building custom AI solutions. Performance is evaluated based on the effective application of AI to achieve results, making consistent and creative use of the latest AI capabilities crucial for success. Candidates must be adept at quickly adopting new models and methodologies, seamlessly integrating AI into their work, and continuously pushing the boundaries of what these technologies can accomplish.
• • We operate at the rapid pace of AI evolution. This means daily work can change quickly, making it an ideal environment for those energized by experimentation, adaptation, critical thinking, and continuous learning. This role may not be suitable for individuals seeking a highly prescriptive, traditional 9-to-5 structure or those hesitant to embrace constant change and learning.
• • The opportunity lies in addressing the fundamental challenges of voice AI. While voice is the most natural human-machine interaction modality, current sequence modeling paradigms struggle to deliver voice AI capable of universal human interaction. These limitations stem from inherent data problems in audio: scarcity, immense diversity across voices, speaking styles, and acoustic conditions, and the prohibitive computational and storage costs associated with high-dimensional data. Deepgram believes that entirely new paradigms for audio AI are essential to overcome these hurdles and democratize voice interaction for everyone.
• • As a Member of the Research Staff, you will be instrumental in defining and executing cutting-edge research initiatives focused on Large Language Models (LLMs). Your role will involve deep dives into the hard technical aspects of LLMs, including sophisticated data curation strategies, efficient distributed large-scale training methodologies, advanced optimization of transformer architectures, and Reinforcement Learning (RL) training techniques.
• • You will be tasked with brainstorming and collaborating with fellow researchers to conceptualize and define new LLM research directions. This includes conducting comprehensive literature surveys, critically evaluating, classifying, and synthesizing current research methods to identify novel approaches.
• • A key responsibility will be designing and executing rigorous experimental programs for LLMs, ensuring that research hypotheses are tested efficiently and effectively. You will drive transformer (LLM) training jobs on distributed compute infrastructure, ensuring successful execution and the subsequent deployment of new models into production environments.
• • Clear and concise documentation and presentation of results, along with complex technical concepts, will be vital to communicate findings effectively to diverse audiences. Staying abreast of the latest advancements in deep learning and LLMs, with a keen focus on their implications and practical applications within Deepgram's products, is paramount.
• • This role is perfect for individuals passionate about AI, eager to contribute to state-of-the-art LLM research, and interested in translating new scientific discoveries into deployable large language models. If you thrive on building systems from the ground up, enjoy creating novel solutions, possess strong communication skills to articulate complex ideas clearly, and are highly analytical with a penchant for detailed investigation, you will find this role exceptionally rewarding.
• • You will be at the forefront of tackling "unsolved" problems, pioneering entirely new approaches in the field. Your ability to identify the single critical experiment that can validate or invalidate an idea rapidly (in days, not months) will be highly valued. Furthermore, you will possess the vision to scale successful proofs-of-concept by orders of magnitude and be driven by an obsession with using AI to automate and amplify your own impact.
• • This position demands an intense focus on the research problems, a creative and innovative approach to solutions, and a relentless drive towards elegant, scalable outcomes. While the technical challenges are immense, the potential to make a transformative impact on the future of voice AI is equally significant. If you are energized by these expectations and already envisioning multiple innovative ideas, you are precisely the researcher Deepgram is seeking.