Dualizer
AI software that turns industrial process diagrams and plant data into engineering simulation models that process engineers can review and optimize.
- SCET Startup Expo · Spring 2026
- AI Venture Velocity Challenge winner
- Renamed Dualizer (formerly EntroPINN)
- Y Combinator Fall 2026 batch
- AVEVA partnership in development

About the project
Dualizer, which showed at SCET under the name EntroPinn, builds physics-grounded AI software for process engineering in industries such as chemicals and energy. Its tools take a plant drawing, operating data or a plain-language goal and turn it into a process simulation model. Engineers can review the model, calibrate it to plant data and use it to test improvements. A conventional simulation engine computes the numbers, not the AI assistant.
The founders, Nathan Liu, Dongha (David) Kim and Christopher Donahue, are UC Berkeley chemical engineering undergraduates with experience in machine learning, controls and process simulation. They presented at the SCET Startup Expo in spring 2026. Their listing described physics-informed optimization tools for industrial heat and process systems and early partnership work with the industrial software company AVEVA.
In September 2026 the team won the $100,000 grand prize in the inaugural AI Venture Velocity Challenge, a national competition run by Texas A&M's Mays Business School that drew 531 teams from 160 institutions. The judges rewarded evidence rather than pitch polish. Over five months the team benchmarked its technology, talked with practicing engineers, narrowed its scope when a broader approach didn't hold up, and improved how it converts diagrams into models. Mays reported that the team next plans to validate its models on new engineering cases and win its first paid industrial pilot. The company's website now lists it in Y Combinator's Fall 2026 batch.
The team
3 members- Nathan LiuChemical Engineering
- Dongha (David) KimChemical Engineering
- Christopher DonahueChemical Engineering
