Analog and RF IC design has traditionally relied heavily on expert knowledge, iterative simulation, and manual layout refinement. Recently, advances in machine learning and large language models have accelerated research toward more autonomous design flows, while industry is increasingly exploring AI-assisted and agentic EDA tools.
This seminar will first provide a brief overview of recent progress in AI-driven analog/RF design automation, from topology generation to physical design and layout synthesis. I will then present two of my recent works: AnaFlow-RF, which combines LLM reasoning, reinforcement learning, surrogate modeling, and EDA tools for specification-to-layout automation, and AnaCLARA, which uses LLM assistance to generate and explain hierarchical analog layout constraints. Together, these works show how AI and LLM-based reasoning can be integrated with optimization and EDA tools to enable more explainable, efficient, and increasingly automated analog/RF design flows.
9/10/2026 11:00 - 12:00
ESAT B91.100
