Bar Lerer

I’m a Lead Software Engineer at Cadence Design Systems, where I build AI for chip verification on the Verisium platform. My work spans regression learning for Verisium AutoFocus, explainable ML for regression data using LightGBM and SHAP, and LLM agents for engineering workflows. One of those agents, an agentic code-review system, won Best Demo at the Cadence Innovation Conference Israel.

Before Cadence, I completed my M.Sc. in Computer Science at Ben-Gurion University of the Negev, researching deep-learning preconditioners for the high-frequency Helmholtz equation. That work, which combines a U-Net CNN with multigrid, was published as a first-author paper in the SIAM Journal on Scientific Computing and presented at an ICLR 2024 workshop. I did my B.Sc. at TU Delft, where I also worked as a research engineer building conversational agents.

I care about taking ideas from research into tools people actually use, and about designing systems carefully before building them: plan first, program second.