Nathan Roll is co-founder and CEO of Oruk, the first company selected for both a16z Alpha and Speedrun. He paused his Stanford Linguistics PhD to build speech foundation models that understand words, emotion, and delivery. A Stanford EDGE Fellow affiliated with Stanford NLP, Nathan works at the intersection of speech AI, computational neuroscience, and model interpretability. His research has appeared in JASA Express Letters, the Journal of Surgical Research, EMNLP, and CoNLL. His first-author EMNLP 2025 paper with Dan Jurafsky and collaborators demonstrated a 19.7% relative reduction in speech-recognition word error rates using roughly 50 seconds of example speech. His collaborative research has received Cambridge Language Sciences funding twice, for work on recognition bias and adapting speech models to neurodivergent speakers. Before Oruk, Nathan founded Strype Analytics and worked as a data scientist at Nalu.ai, which was subsequently acquired. He graduated from UC Santa Barbara with honors and distinction in Economics and Linguistics and earned the department's Top Paper award.