Forki

Forki is a consumer AI app for college students who want to log what they eat without the drudgery of typing every meal into a database. You point the camera at a meal and Forki logs it. The hard part is that "food" means three very different things to a vision model — a packaged product with a barcode, a restaurant dish you didn't prepare, and a homemade meal assembled from whatever was in the fridge. Trying to route all three through the same analyzer is how you end up with GPT-4o telling someone their tikka masala is 400 calories.

I co-founded Forki with Janice Chung and lead product. I designed the AI camera system architecture — a multi-analyzer fallback with three-path routing for packaged, restaurant, and homemade food — to handle the long tail of cases that single-model approaches miss. I co-authored the PRD and character design. Janice led the visual design and brought the app through to production.

We're in private beta with a small cohort of college students. If you're interested in early access or the approach we took, forki.app is the landing page.