SynapseMart — AI-Powered Hybrid Marketplace Search
SynapseMart is an intelligent marketplace search platform that ingests product CSV catalogs and delivers ranked product results through natural-language queries.
Developed as an open-source blueprint under the Cloud2 Labs Innovation Hub, SynapseMart demonstrates how product ingestion, microservices APIs, NLP query parsing, and hybrid retrieval can be combined into a practical end-to-end workflow.
It showcases a production-style microservices architecture with a FastAPI gateway, a Product Service for catalog ingestion and storage, and a Search Service that combines semantic vector search (SentenceTransformers + FAISS) with lexical relevance (BM25).
What It Demonstrates
SynapseMart illustrates how to
- Process product catalogs from CSV uploads without manual data entry
- Automatically deduplicate products during ingestion
- Parse natural-language queries for category and price intent
- Apply hybrid ranking by combining semantic and keyword retrieval
- Filter and sort results using extracted constraints (for example, price under/max and price ordering)
- Keep search indices synchronized by triggering reindexing after catalog updates
Designed for developers, marketplace teams, data engineers, and innovation groups, SynapseMart serves as a reference implementation for AI-enhanced product discovery workflows.
Key Capabilities
CSV Catalog Ingestion Pipeline
Uploads CSV files through the UI and validates, stores, and deduplicates products in SQLite.
Natural-Language Query Parsing
Uses spaCy and rule-based extraction to detect search intent, category hints, and price constraints.
Hybrid Search Engine (FAISS + BM25)
Combines sentence embeddings with lexical scoring and reciprocal-rank fusion to improve relevance.
Category and Price-Aware Retrieval
Applies category-aware boosting and query-derived price filtering/sorting before returning final results.
Gateway-First Microservices Architecture
Routes all frontend requests through a single FastAPI gateway to Product and Search services.
Containerized Development Blueprint
Uses Docker and Docker Compose for reproducible local development and experimentation.