Elman Mansimov is the Founder and Chief Scientist at Ortet, based in New York City. Previously, he was a Senior Applied Scientist at Amazon Web Services, where he worked on agents, foundation models at Amazon Bedrock, and conversational AI at Amazon Lex.
He finished his Ph.D. at New York University where he was supervised by Kyunghyun Cho. His thesis focused on non-autoregressive generation, teaching neural networks to iteratively improve their own rough drafts of text (machine translation) and molecules, years before text diffusion models became a hot topic.
He did an undergraduate degree in Computer Science at University of Toronto supervised by Ruslan Salakhutdinov. During that time, he worked on unsupervised learning from videos and developed the first text-to-image generative model.
For an up-to-date list of publications, please take a look at his Google Scholar.
Highlights
Back in 2015, I created the first generative text-to-image AI model.
The story behind it was covered in this Vox video:
In November 2023, the images from my text-to-image AI model were exhibited at the largest photography fair Paris Photo and Frieze Gallery Space in London.
Several of those pieces have been acquired by the Worcester Art Museum in Massachusetts.
alignDRAW exhibited as part of the New
Terrain: 21st-Century Landscape Photography, Worcester Art Museum (WAM). Photo Credits: Nancy Burns,
Matthew Gamber, and WAM.
November 9, 2023
My work exhibited at the Frieze Gallery Space together with Roope Rainisto's (one of the leading AI artists) interpretation of my prompts.
November 23, 2023
The digital versions of the alignDRAW generated images (NFTs) have been fully minted out (floor price of 0.1 ETH at the time of the mint in December 2023) and are available on OpenSea and other major NFT platforms.
Floor, volume, and other stats of alignDRAW collection
on OpenSea as of end of March 2024.
In September 2026, the collection was added to OpenSea's Flagship Collection, a curated reserve of culturally significant digital art.
September 17, 2026
10 Year Anniversary (2015–2025): To mark the tenth anniversary, I recreated the original alignDRAW model using contemporary hardware while preserving its architecture and training data, and generated new outputs from it. Together with Fellowship, we put together a 291-page hardcover book featuring five essays by Lev Manovich, Shana Lopes, Eva Jäger, Mat Dryhurst, and Darius Himes that explore alignDRAW's place in the history of image-making, from early photography to conceptual art. The project also includes a long-form interview film, dye-sublimation prints made at Griffin Editions in NYC, and more. Learn more →
"alignDRAW images lack photorealistic polish, but they carry the aura of Niépce's first photograph: hesitant, imperfect, epochal." — Darius Himes, International Head of Photographs, Christie's
Software
I like developing SaaS (software-as-a-service) applications in my spare time.
Here are the SaaS applications and services I'm currently highlighting:
- Sourcely — Finish your research in minutes. Simply paste your essay or paper into Sourcely to quickly find, summarize, and add credible academic sources.
- Yomu — AI-powered academic writing tool that removes stress about deadlines. Simplify your writing process with features for editing, paraphrasing, proofreading, and citing sources.
- Revise — Polish your writing without losing your voice. AI-powered editing where you control every change to keep your writing authentic.