Scaling Generative AI: Building Production-Ready LLM Applications
summary
Shows how to build a production LLM app in Java using Quarkus with an MCP tool orchestration layer and EasyRAG for embeddings and retrieval, demoing a working agent that calls web search, memory and Slack integrations. Notes Hugging Face hosts ~2 million models and warns fine‑tuning can require six to ten GPUs with costs in the hundreds of thousands, so recommends RAG, observability, CI/CD and vector stores instead of costly full model training.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:27Generative AI history
- 4:58Generative to agentic AI
- 6:253-step AI adoption
- 9:13Developer AI expectations
- 11:01Quarkus for AI apps
- 14:32Enterprise AI capabilities
- 17:03Live demo: agent AI app
- 28:39AI apps in production
- 30:05Open Data Hub for AI
- 31:09Q&A
related talks
01h 01mScaling AI with Java: Building Production-Ready LLM Applications
48mAgentic AI with Quarkus, LangChain4j and vLLM
47mQuarkus and AI Integrating Java and LLM to build intelligent applications
57mPure Java Enterprise AI/LLM Integration (EAI 2.0)
47mOwn Your LLM: Running Generative and Agentic AI Locally
28mProduction-Ready LLMs on Kubernetes: Patterns, Pitfalls, and Performa...
38mEmpowering AI Education with Rag: Building Domain-Specific LLM Agents for Enhanced...
18mLLMs and GenAI: A Guide to Building Your Own RAG Model for Document Interaction