Bio
Joseph is cofounder/CEO at Roboflow, the vision AI company. Roboflow makes infrastructure millions of AI engineers including half the Fortune 100 use to create and deploy vision models in the cloud and on the edge. They are the authors of RF-DETR, SOTA realtime instance segmentation and detection transformers. Roboflow's backed by investors like YC, GV (fmr Google Ventures), Greg Brockman, Jeff Dean, amongst others.
Sessions (3)
Additional Speaker Details
Using publicly available information we constructed an analysis to help you get a feel for this speaker before deciding to attend their session.
Joseph Nelson brings practitioner depth on production computer vision — why vision still lags language models at scale, how Neural Architecture Search closes the gap, and what it takes to go from dataset to deployed model across manufacturing, healthcare, and robotics. His perspective is grounded in scaling Roboflow's platform to over one million engineers while shipping open-source SOTA research like RF-DETR under real deployment constraints.
Speaking style
Recent talks (1)
GitHub
@josephofiowa- josephofiowa/GA-DSI★ 32General Assembly data science workshop lessons (Python, APIs, data science, visualization)
- josephofiowa/dc-michelin-challenge★ 7Data science competition to predict DC restaurants in the Michelin Guide
- roboflow/rf-detrorg★ 8,200Real-time object detection and segmentation model architecture by Roboflow, SOTA on COCO; accepted at ICLR 2026
Recent writing (3)
Podcasts & interviews (2)
- Training the AIs' Eyes: How Roboflow is Making the Real World Programmable · The Cognitive Revolution · Apr 2026
- Turning Computer Vision Into Real-World Value at Enterprise Scale · The AI in Business Podcast · Apr 2026