A lightweight quick-commerce intelligence platform that consolidates product data across Blinkit, Zepto, and Instamart into a unified product identity model.
The system ingests marketplace data, resolves cross-platform product identities, stores historical pricing and availability snapshots, and exposes a React-based dashboard backed by an Express API.
Quick-commerce platforms often list the same product with different naming conventions, packaging formats, and metadata. This project addresses that challenge by creating a canonical product identity layer that enables meaningful comparison across platforms.
Key capabilities include:
- Cross-platform product matching
- Canonical SKU generation
- Historical price and availability tracking
- Unified product intelligence API
- Interactive React-based product detail dashboard
Products from Blinkit, Zepto, and Instamart are normalized into a single canonical representation using:
- Product type detection
- Flavor and variant extraction
- Weight normalization
- Deterministic SKU generation
Example:
text protein_bar__chocolate_chunk_nuts__60
This allows equivalent products across platforms to be mapped into a single product record while preserving platform-specific listings.
The platform provides:
- Current price comparison
- Availability tracking
- Discount analysis
- Platform-level listing information
- Historical snapshot storage
The ingestion layer:
- Reads marketplace scrape files
- Extracts structured product attributes
- Generates canonical identities
- Stores platform listings
- Records price and availability snapshots
Data Sources:
- Blinkit
- Zepto
- Instamart
- React
- Vite
- Node.js
- Express.js
- Python
- SQLite
- REST APIs
- Canonical Product Modeling
- Snapshot-Based Historical Tracking
text RevQ/ ├── app/ │ ├── frontend (React + Vite) │ └── backend (Express API) │ ├── ingest/ │ └── Python ingestion pipeline │ ├── data/ │ ├── blinkit_sample.json │ ├── zepto_sample.json │ └── instamart_sample.json │ └── database/ └── SQLite
https://quickcommerceintelligencesystem-mlbk8ybw7mhtgicvdse7qo.streamlit.app
Different platforms frequently represent identical products with slightly different naming conventions.
By separating:
- Canonical Products
- Platform Listings
the system can support accurate comparisons while preserving source-specific information.
SQLite provides:
- Zero configuration
- Fast local development
- Simple portability
making it suitable for rapid prototyping and take-home exercises.
The matching logic is intentionally transparent and explainable.
Rather than relying on opaque heuristics, normalization rules are visible, auditable, and easy to improve over time.
- Matching rules are handcrafted for the sample dataset.
- Pack count is not modeled separately.
- Historical trend visualization is not yet implemented.
- API does not include authentication or pagination.
- SQLite access is optimized for local development rather than production workloads.
- Confidence scoring for product matching
- Human review workflow for ambiguous products
- Price history visualizations
- Automated data quality validation
- Materialized latest-price views
- Comprehensive unit and integration testing
- Production-grade database layer
Aayushi Pandey
Full-Stack Developer & AI Engineer passionate about building products that don't just live in repositories—they solve real-world problems and create meaningful value for users.