← All work
DOMAIN LOGIC / REASONING2024–2025

AI
Nutritionist

Expert meal-planning judgment, translated into rules.

ENTERPRISE AI / CUSTOMER PROJECT

Case study by

Meal-planning interface and configurable business rules
Meal-planning interface and configurable business rules

The problem.
My part in it.

Background

  • Menu planning combined nutrition, budget, supply and kitchen limits.
  • Ingredient changes triggered substitution and cost reviews.
  • Experienced dietitians held much of the required judgment.
  • That knowledge needed reuse across sites and junior staff.

Objective

  • Reduce repeated planning and review work.
  • Share expertise through recommendations suited to each site.

My role

  • Joint discovery of the business need and planning problems.
  • Structured data and rules; aligned scope and acceptance criteria.

How it connects.

A connected system for constrained meal planningSix modules connect enterprise data, ingredient reasoning and nutritionist decisionsA connected system for constrained meal planningSix modules connect enterprise data, ingredient reasoning and nutritionist decisionsINPUT / ENTERPRISE DATA + LOCATION SETTINGSINTELLIGENCE / DATA → LOCATION-SPECIFIC CANDIDATESPLANNING / REASON ABOUT COMBINATIONS AND CONFLICTSOUTPUT / NUTRITIONIST REVIEW & ENTERPRISE INTEGRATIONMaster data — Ingredients / recipes / pricesMaster dataIngredients / recipes / pricesOperating history — Orders / servings / preferencesOperating historyOrders / servings / preferencesPlanning request — Calendar / budget / menu rulesPlanning requestCalendar / budget / menu rules01 RI classifier — Raw names → standard hierarchy / Units / packaging / attributes01 RI classifierRaw names → standard hierarchyUnits / packaging / attributes02 Menu enrichment — Cooking method / taste / texture / Main ingredient / country02 Menu enrichmentCooking method / taste / textureMain ingredient / country03 Ingredient optimizer — Availability + rules + ranking / Local recipes, cost, nutrients03 Ingredient optimizerAvailability + rules + rankingLocal recipes, cost, nutrients04 Menu-combination AI — Slots / variety / compatibility / Menu candidate pool04 Menu-combination AISlots / variety / compatibilityMenu candidate pool05 Schedule generation — Day → meal → corner → serving / Budget / frequency / nutrition05 Schedule generationDay → meal → corner → servingBudget / frequency / nutrition06 Rule-conflict engine — Constraint detection / Priority / adjustment06 Rule-conflict engineConstraint detectionPriority / adjustmentMeal plan + ingredient choices + cost + nutrients + rule compliance
Processing / data flowIteration / feedbackService / environment boundary
01

Hard constraints and preferences.

Apply prohibitions and supply conditions to the candidate set, then use site preferences in recommendations. Make priorities and exceptions editable by administrators.

02

Six modules, explicit responsibilities.

Separate classification, enrichment, ingredient optimization, menu composition, period planning and conflict resolution. Align output expectations and acceptance criteria with the customer and engineering team.

Business-rule configuration

The outcome.

−80%Meal-planning time
95%Ingredient recommendation accuracy
99%Within-budget meal plans

Figures reported in the project presentation. My contribution: business rules, data and slots, scope and acceptance criteria.

NEXT PROJECT / 03Utility Chatbot