Real-time AI Lakehouse
summary
Claims a working real-time AI lakehouse built on HopWorks: code uses the BestTime API for foot-traffic, HopWorks feature store, model registry and KServe-style serving, LangChain/OpenAI for LLM function-calling, XGBoost for the baseline model and optional embedding nearest-neighbors. The tutorial assembles feature, training and inference pipelines into an MVP recommendation (find a bar in an hour), runs end-to-end in about half an hour on the free tier but hits API/credit limits, and shows practical gaps around orchestration and production scaling of feature store + model registry and LLM context retrieval.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 4:45Operationalizing ML challenges
- 9:18Simplifying ML pipelines
- 13:57MVP for bar recommendations
- 20:40Code walkthrough
- 37:43Training and registering model
- 51:35Inference pipeline
- 1:03:13Orchestration complexity
- 1:08:52Resources and closing
related talks
30mPython and the AI Lakehouse
29mLakeside AI: Bringing Models to Data with Open Lakehouses & Agents)
05mBackstage interview
49mBuild & deploy generative AI apps in days with Databricks Lakehouse AI
23mBuilding AI-Driven Real-Time Lakehouse APIs
26mReproducible AI with Kubeflow, lakeFS and Langchain
58mAI-Enhanced Big Data: Integrating Private LLMs and Vector Databases
19mReal-Time AI: Low Latency Solutions for Interactive Applications