Are LLM based Recommendation Systems a Model Problem or a UX Problem?
Google meet link: https://meet.google.com/itx-spmw-mmy
Recommendation systems have long driven revenue for consumer-focused companies, evolving from classical ML to deep learning models. While traditional methods struggled with long-term context, seasonal patterns, cold-start problems, and evolving user behaviour, LLMs have made rich context embedding more feasible. Yet, their scalability challenges in real-time raise a critical question: how can we best integrate them into modern RecSys? Should LLMs generate offline embeddings, enhance search, or act as conversational agents for hyper-personalized recommendations? In this talk, I’ll explore how advances in models and UX are reshaping the future of AI-driven recommendations.
About the speaker
Anindyadeep Sannigrahi
Current: ML Engineer @PremAI