Building Multi-modal Applications with a Real-time Database | Real-time Data Summit 2024

Learn about the construction of multi-model applications using Aerospike. Try Aerospike Cloud for free: https://auth.control.aerospike.cloud/...

In this talk, Art Anderson, Senior Developer Experience Engineer at Aerospike, walks through building a multi-model application using the Aerospike database. Discover how Aerospike seamlessly supports key-value, document, graph, and vector data models all in a single, scalable platform.

🚀 What You'll Learn:

How Aerospike enables real-time, sub-millisecond data access at scale

Using Aerospike Vector Store for semantic search and product similarity

Leveraging Aerospike Graph Service to traverse customer-product relationships

Building a recommendation engine using vector similarity and graph traversal

Creating a rich user experience with natural language search on product images and text

The advantages of a multi-model vs. single-model database architecture

A full demo using a retail website with 44,000+ fashion items and 100,000+ synthetic customer profiles

🛒 Real-World Demo Includes:

Key-value lookups for individual products

Semantic vector search using OpenAI’s CLIP model for natural language and image-based search

Graph-based insights like “also bought” recommendations

Use of a unified Aerospike namespace for tight integration between key-value and vector data

Efficient data management, fewer moving parts, and simplified operations

👨‍💻 Art shares practical tips for developers, architects, and data engineers interested in:

Combining vector databases, graph databases, and key-value stores

Running recommendation systems and search applications

Managing real-time e-commerce platforms at scale

Using AI-powered embeddings and semantic search in production

📦 Tech Stack Highlights:

Aerospike Database

Aerospike Vector Search (HNSW, cosine similarity, dot product, etc.)

Aerospike Graph (Apache TinkerPop / Gremlin interface)

OpenAI CLIP model for generating vector embeddings

Synthetic data from Kaggle for retail product and customer profiles

Whether you're building AI-powered apps, recommendation engines, or knowledge graphs, this session showcases how Aerospike’s multi-model capabilities can streamline architecture and reduce operational complexity all while delivering lightning-fast performance.

Like, comment, and subscribe to stay up to date on developer content, product demos, and technical deep dives from the Aerospike team.

Explore Aerospike: https://www.aerospike.com

Aerospike Database
https://aerospike.com/products/database/

Aerospike Documentation Guide
https://aerospike.com/docs/

Build with Aerospike
https://aerospike.com/docs/develop/

#Aerospike #VectorSearch #GraphDatabase