Babak Barazandeh

WORK HISTORY

EDUCATION

ABOUT BABAK BARAZANDEH

As a Staff Data Scientist at Fortinet, I design and deploy advanced Generative AI and Agentic AI capabilities, blending large-scale optimization with modern LLM architectures. I led the development of FortiAI, Fortinet’s LLM-driven intelligent assistant, and introduced Localized LoRA, a structured fine-tuning method that significantly improves stability, efficiency, and performance when adapting large language models to real-world enterprise tasks. I also develop stable, optimization-grounded fine-tuning algorithms that enable LLMs to act as reliable autonomous agents—capable of analyzing, planning, and making context-aware decisions.My earlier research on Generative Adversarial Networks (GANs) resulted in optimization techniques that dramatically improved training stability and reduced computational cost, earning a Best Paper Award at the Data Science Workshop. These contributions form part of the foundation I now apply to large-scale LLM adaptation.Previously at Splunk, I led the technical development of Splunk Copilot, a GenAI system that translates natural language into SPL queries, accelerating analytical workflows and enabling automated reasoning over machine data. Across Fortinet, Splunk, and the University of Florida, I have built deep expertise in generative modeling, non-convex optimization, and adaptive, self-improving AI systems. My mission is to advance the practical foundations of GenAI and Agent AI by transforming research breakthroughs into scalable, high-impact intelligent systems.