JavaOne,JavaLand, GenAI, Langchain4J, Lombok, Rolling Updates, Sync/Async, Serverless--133rd airhacks.tv 📎
"Treating LLMs as microservices with APIs for enterprise integration, Java is excellent for model inference and production deployment, Enterprise applications benefit from LLMs as decision support tools rather than decision makers, Langchain4j gaining adoption in the Quarkus enterprise space, Java developers and Python developers form distinct groups with different priorities, Future trend toward smaller dedicated models through model distillation, Java's vector API can be as fast as modest GPUs in some cases, Local LLM inference possible with Java at 10 tokens per second, Java 24 released with gatherers and FFM API, Java One conference was well-organized with great content and networking, JavaLand conference focused on community with keynote about 30 years of Java, Centralized logging now standard in cloud environments, Application-scoped services preferred in serverless architectures, Generic utility packages should be avoided for better cohesion, Preference for JSON-P over JSON-B for better versioning capabilities, Synchronous communication preferred over queuing when possible, Vaadin making a comeback with web components support"...is ready to watch:
See you every first Monday of the month at https://airhacks.tv 8pm CET (UTC+1:00). Show is also announced at: meetup.com/airhacks.
Are you nice? :-) Then checkout: the airhacks discord server