Harrison Vanderbyl is Co-Founder and CTO at Featherless AI, a serverless inference platform on a mission to democratize access to the world's largest collection of open-source AI models.

Based in New South Wales, Australia, Harrison emerged from the RWKV open-source community and co-founded Recursal AI, where he served as CTO and led the development of next-generation linear attention AI architectures.

Harrison is a key contributor to RWKV, the first AI model architecture under the Linux Foundation, and co-authored research on Gated Linear Attention for neural speech synthesis, demonstrating deep expertise in AI infrastructure and model design.

At Recursal AI, Harrison led training of the EagleX model, achieving 7,500 tokens per second on AWS Inferentia using just 1% of the infrastructure footprint of traditional transformer architectures.

Featherless AI has raised $25M in total funding including a $20M Series A co-led by AMD Ventures and Airbus Ventures, and now serves over 30,000 open-source models as Hugging Face's fastest-growing inference partner.