Build a personalized Commute virtual assistant in Python w/ Hopsworks & LLM Function Calling
summary
Builds a working Python commute assistant that uses LLM Function Calling to decide whether to invoke two concrete functions: a predictor that queries a Hopsworks feature store via a Feature View and a history fetcher that calls the real-time SL API and a Kafka stream. Demo shows low-latency probability predictions served from a deployed model, end‑to‑end retrieval + LLM responses, some token/context limits and occasional function misclassification but overall executable and reproducible.
AI-written summary · about
chapters (16)
- 0:00Personalized commute assistant intro
- 1:10LLM function calling limits
- 6:12Hopsworks & FDI pipelines
- 9:05Commute assistant architecture
- 12:18Demo setup in Hopsworks UI
- 14:12Streaming feature pipelines
- 16:20Delay prediction training
- 17:44Inference with function calling
- 21:32LLM prompt templates
- 24:01Live commute assistant demo
- 27:59Handling departure queries
- 29:24Hopsworks community resources
- 30:12Q&A function calling
- 32:21Q&A prompt design
- 34:49Q&A context & features
- 36:39Q&A code & tutorials
related talks
29mBuild a personalized Commute agent in Python with Hopsworks, LangGraph and LLM Function Calling
47mBuilding Python based AI Systems with LLMs
38mImprove LLM-based Applications with Fallback Mechanisms
01h 10mReal-time AI Lakehouse
31mHow to build a serverless electricity price prediction service in Python with Hopsworks + Streamlit
29mLLM-Based Development
01h 23mBuild a production ML system with only Python on free serverless services
41mBuilding LLM Agents – Challenges and Learnings