Built by Atlanta IT specialist Mark Johnson, CompKitchen uses scientific databases and molecular flavor profiles to help home cooks and food scientists discover unexpected ingredient pairings.
Atlanta founder and IT specialist Mark Johnson built CompKitchen after years of wondering how great cooks (like the ones he watched on MasterChef) seemed to instinctively know which ingredients belonged together. Now, his food pairing platform uses molecular flavor chemistry, research papers, and scientific databases to help home cooks and food scientists discover why certain combinations work.
Johnson has been building CompKitchen on nights and weekends, bootstrapping the platform that surfaces flavor relationships that home chefs and food scientists alike can use. CompKitchen’s platform is built on 30,000 research papers and 11 scientific databases to help people see why certain ingredient combinations work at the molecular level. That means that users can move beyond trial-and-error cooking and explore unique pairings based on shared compounds, flavor chemistry, and scientific evidence.
Foodies on the internet have taken note of the FoodTech startup.
The product launched on Product Hunt. In its early stages, it has grown to about 11,000 monthly visitors with no paid marketing spend behind it. A viral Reddit post put the CompKitchen platform in front of 180,000 more eyeballs.
How is CompKitchen different from AI recipe tools?
AI has, of course, irreversibly changed how we search for information in digital spaces. But Johnson said that CompKitchen works in a way that traditional LLMs do not.
“The platform is grounded in measured chemistry, not language model inference. There’s a real difference between asking an LLM ‘what pairs with peach?’ and querying a database of volatile compound profiles extracted from 40,000+ peer-reviewed papers and 11 scientific databases. The first gives you plausible-sounding suggestions based on patterns in text. The second reports what’s actually been measured,” he told Hypepotamus. “The methodology is also published and validated. There’s a preprint on Zenodo (indexed with OpenAIRE) with quantitative cluster validation, and I’m running bench experiments to test the system’s predictions against actual cooking. That’s a different kind of proof than ‘the AI suggested it.’”
The platform offers free, pro, enterprise, and academic pricing plans.
Meet The CompKitchen Founder
By day, Johnson is a Technical Support Specialist II at Fulton County Schools.
Scaling CompKitchen on the side has been hard, he admitted to Hypepotamus, saying it requires a lot of “pre-dawn sessions and weekend bench tests.”
He’s bootstrapping the platform in its early stages. “The structure is intentional,” he told Hypepotamus. “The day job covers my expenses, which means CompKitchen doesn’t have to make money tomorrow. There’s no investor clock, no runway pressure, no rush to ship features that won’t make the product better. I can work on the right problems on the right timeline. For something this technical, that matters more than speed.”
So we had to ask Johnson: What have been some of the most interesting flavor profiles he has discovered through the platform so far?
His answer: passionfruit mixed with chamomile.
“They share 108 volatile compounds but appear in literally none of the 600,000-recipe corpus I’ve indexed. A tropical fruit and an aromatic herb with no obvious culinary relationship, but molecularly they’re closely related,” he shared with Hypepotamus. “That’s the kind of finding I built the platform to surface.”
As he works on building CompKitchen, Johnson said there is a bigger, less quantifiable goal driving him.
“There’s a generation of home cooks who’ve drifted away from actually making things, partly because the gap between a recipe and a great result is bigger than recipes admit. CompKitchen exists to close that gap by exposing the chemistry. If I can get more people creating again, that’s the real win.”