MealPalBot
A plug-in bot + web UI that generates retrieval-grounded meal plans using OpenAI + RAG and converts them into a checkout-ready grocery cart flow.
Overview
MealPalBot is a plugin-style assistant (A bot with an optional web UI) that orchestrates meal planning from preference + pantry inputs and turns the output into an actionable grocery ordering workflow.
The Problem
Meal planning isn’t just choosing recipes,users still have to translate preferences, pantry context, and nutrition constraints into a concrete ingredient list and then manually build a cart/checkout flow. That last mile creates most of the friction and drop-off.
The Solution
Implemented a split-service architecture: Django MVC + DRF acts as the system of record for users, recipes/nutrition, pantry state, meal-plan objects, and OCR ingestion. A NestJS service handles OpenAI-based meal planning with a RAG layer over the Django-managed dataset and owns Kroger OAuth + cart integration to generate purchasable carts. Added Redis caching and geolocation-aware store selection surfaced in the Next.js UI to reduce latency and improve local relevance.
Results & Impact
Built an end-to-end pipeline from user inputs → retrieval-grounded meal plan → normalized ingredients → cart-ready items, with faster repeat responses via caching and improved usability through OCR-based ingredient capture.