Yicheng Guo

Modern engineering teams have world-class observability for apps, infra, and data. Yet when it comes to voice, most teams are still operating in the dark (before they come to us at Hamming AI). Latency spikes, broken prompts, ASR drift, mispronunciations, etc, directly shape the customer experience, but without proper visibility, teams only discover problems after customers churn. That’s why I believe voice observability must emerge as its own discipline, complete with its own metrics, frameworks, and industry standards. Read more here: https://lnkd.in/eTTCUCQp

46

2 Comments

Exciting news! AWS just launched S3 Vectors—native, scalable vector search right in S3! Store and query massive AI datasets with subsecond latency and lower costs. Game-changing for AI, RAG, and vector workloads.

51

2 Comments

GPT-5 has arrived, and it's live across M365 Copilot, GitHub Copilot, Copilot, and Azure AI Foundry! It's the most powerful new LLM across several benchmarks and everything runs on Azure, with Foundry providing the coordination and security to support enterprise use. Azure AI Foundry's endpoint using model router allows developers to automatically direct their prompts to the most suitable version — whether it’s GPT-5, GPT-5-mini, GPT-5-nano, or GPT-5-chat. https://aka.ms/GPT-5-blog

70

4 Comments

"Build a Data Lakehouse with Apache Iceberg, Polaris, Trino & MinIO" Enjoyed reading this two-part blog post series by Gilles Philippart, teaching how to run a Data Lakehouse locally, using open-source technologies: - MinIO for object storage - Apache Iceberg as the open table format - Apache Polaris as the metadata catalog - Trino as the query engine. The second part adds Apache Kafka and Flink for streaming data ingest. I also learned about JR, a CLI utility for creating randomized Kafka test data. Everything running locally on your machine via Docker Compose, which is great for learning and experimenting. Really neat, definitely recommend to check it out! 👉 https://lnkd.in/e73fSnPS

201

13 Comments

The Future of #DataStreaming with Apache Flink for #AgenticAI => My latest blog post: Agentic AI is moving from hype to production. These aren’t just reactive assistants. They are autonomous, goal-driven systems that need to operate on live data, interact with tools, and collaborate across protocols like #MCP and #A2A. In my latest blog post, I explore why Data Streaming with #ApacheKafka and #ApacheFlink provides the backbone for scaling Agentic AI. With Flink Agents (FLIP-531), Flink is evolving into a native runtime for always-on, system-triggered AI agents that can maintain memory, make real-time decisions, and integrate seamlessly with enterprise environments. This shift marks a turning point: moving beyond demos and prototypes toward enterprise-ready, event-driven #AgenticAI. It’s the same playbook that turned #microservices from theory into scalable production systems—powered by a robust, decoupled backbone. The future of AI is not batch-driven or manually orchestrated. It’s event-driven, contextual, and continuous. And Flink is positioning itself at the center of this transformation. https://lnkd.in/e4D7rb7h

45

3 Comments