The AI hype is relentless: Autonomous Agents are taking over, and “Classic” Data Science is supposedly dead. This FOMO is driving a dangerous architectural trap where heavy generative models are used as the default solution for everything. But forcing an LLM to scan petabyte-scale data for anomalies isn’t just inefficient- it’s the wrong mathematical paradigm. Falling into this “Tokenization Trap” destroys the data’s natural topology, trading the deterministic bounds required for critical pipelines for probabilistic, hallucinated guesses.
In this session, we’ll ditch the hype and explore “Cascading AI”- a hybrid architecture pairing the deterministic precision of classic ML with the reasoning power of GenAI. Think of it as building a “Nervous System” and a “Brain.” We’ll show you how to deploy lightweight statistical gatekeepers directly inside your compute engines to isolate anomalies with sub-millisecond latency. You will walk away with reusable blueprints for “Semantic Translation”- compressing data anomalies into precise metadata payloads that trigger your LLM “Brain” for deep Root Cause Analysis.
Rotem is a Senior AI Scientist at Intuit, specializing in building large-scale processes that support customers and produce measurable results. Her areas of expertise include Generative AI LLMs, Classical ML, and NLP.
She holds both an M.Sc. and a B.Sc. in Industrial Engineering from Tel Aviv University.
Rotem is dedicated to supporting various initiatives, such as women’s empowerment, serving as the Israeli ambassador for Women in Data Science (WiDS).דשן
Keynote session: Hadas Grossmon Ella
Break
Lightning talks session
Roundtable closing
Talk by Hila Paz
Talk by Dr. Moran Mizrahi
Closing remarks
End
