Generative Menu Engineering
Vision LLM + real-time menu analytics engine for Haiba Bistro. Optimizes dish placement, pricing, and visual hierarchy to boost average order value by 42%.
The Context
Bistro menus are historically static and decoupled from actual daily ingredient margins, waste spikes, or real-time customer purchasing habits.
We wanted to treat the physical restaurant menu as a dynamic, mutating landing page that self-optimizes in response to supply chain and operational inputs.
What I built
- Vision LLM menu analysis. Engineered a Python loop utilizing Vision LLM models to analyze visual menus, proposing layout mutational features to highlight high-margin dishes.
- Waste-to-margin pipeline. Hooked inventory data directly into the menu generator, dynamically pushing inventory-heavy dishes (like seasonal specials) when stocks peaked to cut ingredient waste by 15%.
- A/B layout testing. Deployed alternating weekly menu layouts, tracking margin differences through Looker dashboard integrations.
By the numbers.
If we worked together.
I bring high-leverage product engineering and absolute operational discipline: shipping weekly, pressure-testing user inputs, and automating delivery pipelines.
Jamahook
EU music-tech platform — full Stripe checkout for 15,000+ active users with GDPR + VAT compliance. Component system in Storybook keeps the front-end coherent across teams.