Aditya Ramesh
VP of Research at OpenAI who leads Worldsim; first author of DALL·E and DALL·E 2, co-author of CLIP
Aditya Ramesh leads Worldsim as a VP of Research at OpenAI and, according to Ramesh's homepage, is working on a new robotics effort that brings video-generation models to the physical world. Ramesh is first author of the papers behind DALL·E (2021) and DALL·E 2 (2022), led the team that created DALL·E 3, built the team that created the first Sora model, and worked on controllability and language grounding for Sora 2. Ramesh is also a co-author of the CLIP paper, written with an OpenAI affiliation.1234
Work
CLIPOpen weights
Co-author of "Learning Transferable Visual Models From Natural Language Supervision" (2021), which releases code and pre-trained weights; the repository is under the MIT license.45
DALL·E (external site: arxiv.org)Published research
First author of "Zero-Shot Text-to-Image Generation" (2021), which models text and image tokens as a single stream with a transformer; Ramesh's homepage describes creating this first DALL·E model.21
DALL·E 2 (external site: arxiv.org)Published research
First author of "Hierarchical Text-Conditional Image Generation with CLIP Latents" (2022); the homepage describes co-creating DALL·E 2 with inpainting, variations and blending.31
DALL·E 3, Sora and Sora 2Proprietary product
Ramesh's homepage says Ramesh led the team that created DALL·E 3 (2023), built the team that created Sora (2024), and focused on controllability and language grounding for Sora 2 (2025).1
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.