Teknium leads post-training at Nous Research and builds Hermes Agent, an open-source agent harness, alongside work on the Hermes model family, datasets, and reinforcement-learning environments. They co-authored the Hermes 3 and Hermes 4 technical reports and previously contributed research and dataset preparation to CarperAI/Stability AI’s StableBeluga models.
Co-Founder, Post-Training Lead and Hermes Agent Engineer
Builds open-source language models, agents, datasets, synthetic data pipelines, and reinforcement-learning environments.
Research Contributor
Contributed research, ablation planning, and dataset cleaning for StableBeluga / Free Willy models; a research collaboration.
