Research Staff, LLMs

Deepgram is a leading platform for the Voice AI economy, providing real-time APIs for speech-to-text (STT), text-to-speech (TTS), and production-grade voice agents. Over 200,000 developers and 1,300+ organizations build voice offerings powered by Deepgram, including Twilio, Cloudflare, Sierra, Decagon, Vapi, Daily, Cresta, Granola, and Jack in the Box. Deepgram's voice-native foundation models are accessed via cloud APIs or self-hosted/on-premises software, with a focus on accuracy, low latency, and cost efficiency. Backed by a recent Series C, Deepgram has processed over 50,000 years of audio and transcribed more than 1 trillion words.

Deepgram is seeking an experienced researcher with extensive work on Large Language Models (LLMs) and a deep understanding of transformer architecture to join its Research Staff. The role involves tackling hard technical aspects of LLMs, such as data curation, distributed large-scale training, optimization of transformer architecture, and Reinforcement Learning (RL) training.

• Brainstorm and collaborate with other Research Staff to define new LLM research initiatives.
• Broadly survey literature, evaluating, classifying, and distilling current methods.
• Design and carry out experimental programs for LLMs.
• Drive transformer (LLM) training jobs on distributed compute infrastructure and deploy new models into production.
• Document and present results and complex technical concepts clearly for target audiences.
• Stay up to date with the latest advances in deep learning and LLMs, particularly their implications for Deepgram's products.

• 3+ years of experience in applied deep learning research, with a solid understanding of neural network types, architectures, and loss mechanisms.
• Proven experience working with LLMs, including data curation, distributed large-scale training, optimization of transformer architecture, and RL training.
• Strong experience coding in Python and working with PyTorch.
• Experience with various transformer architectures (auto-regressive, sequence-to-sequence, etc.).
• Experience with distributed computing and large-scale data processing.
• Prior experience in conducting experimental programs and using results to optimize models.
• Preferred: Deep understanding of transformers, causal LMs, and underlying architecture; understanding of distributed training and inference schemes for LLMs; familiarity with RLHF labeling and training pipelines; up-to-date knowledge of recent LLM techniques; published papers in deep learning research, particularly related to LLMs.

Estimated base salary $150K – $250K, with equity and a 10% annual bonus.

Apply via the provided application link.