Research Engineer, Post-Training

At Deepgram, a leader in Voice AI technology, the research team focuses on advancing speech-to-text, text-to-speech, and voice agent capabilities. The company supports a vast developer community and enterprise clients with highly accurate, low-latency voice AI models accessible via cloud or on-premises solutions. The team operates with an AI-first mindset, rapidly adapting to new AI tools and innovations to maintain their cutting-edge position in the voice technology space.

The Research Engineer role centers on identifying and resolving limitations in Deepgram's voice AI models by translating customer challenges into measurable improvements. Daily tasks include designing evaluations, fine-tuning models, generating data, and quickly deploying solutions to production. The position demands creativity, agility, and a strong customer focus to deliver impactful fixes that enhance voice AI performance across diverse applications and languages.

Ideal candidates have extensive experience with deep learning model training and evaluation, proficiency in Python and PyTorch, and a proven ability to ship machine learning solutions in production environments. Familiarity with speech technologies and distributed training is advantageous. The role is fully remote within the USA, requiring adaptability to a fast-paced, evolving AI landscape and a commitment to continuous learning and experimentation.