MXP Platform
AI-powered search and personalization accelerator for e-commerce
MXP is a composable search and personalization platform built by Grid Dynamics CTO R&D. It bundles semantic search, vector search, merchandising rules, AI-driven catalog enrichment, and real-time recommendations into a single deployable stack on GKE.
It is designed to be forked and customized for individual customers, integrating Google Retail API, Vertex AI, Gemini AI, Elasticsearch/OpenSearch, and custom ML models.
Platform capabilities
| Capability | What it does |
|---|---|
| Product Discovery | Semantic and full-text search with multi-stage query processing, facets, and fallback strategies |
| Query Understanding | Parses natural language queries into semantic graphs for intent extraction |
| Merchandising rules | Business-driven boosting, burying, and filtering of search results |
| Catalog enrichment | Gemini AI-powered attribute generation for sparse product catalogs |
| Recommendations | ML-based recommendation containers served via two FastAPI microservices |
| Conversational search | LLM-driven multi-turn search experience |
| Visual search | Image-to-product search via Vision API and vector embeddings |
| Merchandising admin UI | Angular web app for managing rules, facets, and user groups |
Where to start
- New to the platform? → Getting started
- Looking for how things fit together? → Architecture overview
- Exploring features? → Features
- Looking for solutions? → Solutions