Cracking the Nut, Solving Edge AI with Apache Tools and Frameworks
summary
A concrete pipeline shows devices and Jetson gateways ingesting sensor and camera data with MiNiFi → NiFi → Pulsar, running on-device or gateway inference with MXNet/DJL inference on Jetson, streaming results into Flink for realtime queries and joins via Flink SQL over Pulsar topics, and persisting to sinks like Postgres or Apache Kudu using Pulsar connectors and JDBC sync. Demos use Go and Python producers, NiFi processors, Pulsar functions, and demonstrate Pulsar geo‑replication and multi‑protocol (MQTT/JMS/Kafka) support.
AI-written summary · about
chapters (5)
- 0:00Start
- 6:52Who's the Flip Stack Good for
- 7:51Deep Learning
- 9:05Reason Why I Suggest Apache Pulsar
- 16:17Demo Walkthrough
related talks
41mAdding Generative AI to Real Time Streaming Pipelines
32mFrom Edge To Insight: Cybersecurity Intelligence With Kubeflow...
28mBuilding a Coud Native Data Pipeline for Your Real-Time ML & AI Apps
55mLarge Scale Distributed AI on the JVM
26mBuilding a Cloud Native Edge Platform for Accelerating AI Inference
25mAccelerating Data Processing and Consumption for Building Modern Apps at the Edge
34mBuild ML Enhanced Event Streaming Apps with Python Microservices
49mData pipelines: building an efficient instrument to make custom workflow