其他行业 · AI应用
利用DataStax Astra DB和AWS将用户参与度提升并延迟降低90%
Arré Voice面临通过个性化内容提升用户参与度同时高效管理大量数据的挑战。为此，他们基于微服务架构开发了一个推荐引擎，利用DataStax Astra DB和AWS。这一创新方法通过AI驱动算法处理用户数据，每月提供超过3000万条个性化推荐。该实施不仅实现了90%的延迟降低和99.9%的可用性，还使开发成本降低了45%，最终改变了用户体验和运营效率。
Monthly personalized recommendations
30 million
来源披露
Latency reduction
90%
来源披露
Availability
99.9%
来源披露
01
业务背景
Arré Voice面临通过个性化内容提升用户参与度同时高效管理大量数据的挑战。为此，他们基于微服务架构开发了一个推荐引擎，利用DataStax Astra DB和AWS。这一创新方法通过AI驱动算法处理用户数据，每月提供超过3000万条个性化推荐。该实施不仅实现了90%的延迟降低和99.9%的可用性，还使开发成本降低了45%，最终改变了用户体验和运营效率。
02
遇到的问题
Arré Voice wanted to enhance user engagement through personalized content recommendations but faced challenges with its monolithic architecture, which made it difficult to scale and innovate quickly. The initial cloud-based database was costly, had a major learning curve, and lacked integration capabilities for AI-driven algorithms.
03
AI 解决方案
Transitioned to a microservices architecture with 25 microservices orchestrated by Amazon ECS. Selected DataStax Astra DB as the core vector database for real-time queries and low latency. Integrated Amazon DynamoDB for high-throughput operations, Amazon Neptune for relationship mapping, Amazon OpenSearch Service for full-text search, Amazon Bedrock and Amazon Titan for speech-to-text and translation in multiple Indian languages, and AWS Batch for training ML models. This setup enabled real-time personalized recommendations across multiple languages.
04
实施结果
Monthly personalized recommendations：30 million
Latency reduction：90%（reduced to ~100 ms）
Availability：99.9%
Development cost savings：45%
05
风险与边界
未披露
“Our recommendation engine surpassed our expectations, delivering 30 million personalized suggestions each month.”
Shabeer Muhammed · Senior Software Engineer at Arré Voice
信息来源
1 个来源页面内容为结构化改写；关键结论应可追溯到以下材料。