Lessons Learned from Building a Managed RAG Solution

Retrieval Augmented Generation has gained attention in the space as a way to reduce LLM hallucinations and fine tuning costs, while improving product flexibility. In this talk we'll discuss the technical and business lessons learned from building a managed RAG solution into Voiceflow (the Knowledge base). We'll dive into challenges like benchmarking, testing and supporting flexible workflows. The talk assumes basic knowledge of LLMs, Vector DBs and Chatbots.

Speaker:
Denys is head of Machine Learning at Voiceflow focused on building performant real time ML systems. He is a strong advocate of building practical and reliable ML systems in both the Gen AI and more traditional NLU space. He is an active member of the ML community leading discussion groups, mentorship sessions and answering questions on forums. He previously worked as a Senior Cloud Architect at a Global Bank.