2025-06: Mind the RAG Gap: Lessons from Production Chatbots

Think building a RAG chatbot is as easy as uploading some PDFs? Think again! Join us to uncover the real-world pitfalls and principles of shipping production-ready AI chatbots, presented by Saurabh Anand.

Where: Platform Calgary, East Annex
When: Wednesday, June 25, at 5:30pm

Mind the RAG Gap: Lessons from Production Chatbots
Retrieval-Augmented Generation (RAG) looks easy. Drop your docs into a vector DB, fling them at an LLM, and ship. Reality hits harder when you're building a production-grade RAG chatbot: custom chunking strategies, vector-DB quirks, bi-encoder vs. cross-encoder trade-offs, trimming with Maximum Marginal Retrieval, context-window illusions, metadata filters, and the “lost-in-the-middle” curse that buries your gold paragraph. This session will be straight to the point: everything broken down to first principles, sharing practical lessons and techniques of moving a RAG chatbot into production.

Schedule:
5:30 - Food and Networking
6:00 - Presentation and Discussion
7:30 - Wrap up

Bio:
Saurabh Anand is a Data Scientist at the Office of Advancement at the University of Calgary, where he works on predictive models, AI and enterprise-level RAG systems. Previously he delivered data science solutions in banking and oil & gas sectors. Outside work he volunteers in Calgary’s tech community and runs a YouTube channel - contributing and creating open source projects addressing community needs, breaking down research papers, coding tutorials, and exploring Vedanta and philosophy.