About Dai Le

Dai works across software development, frontier-model evaluation and benchmark design, biological research, and scientific computing. His AI work focuses on designing, reviewing, and stress-testing scientific and technical evaluations for advanced models and coding agents, including task specifications, evidence design, scoring and verification, adversarial testing, difficulty calibration, and failure analysis. His research background includes bacterial physiology, antibiotic response, drug-target kinetics, proton motive force, efflux-mediated drug resistance, gene expression under stress, retinal developmental biology, retinal pigment epithelium polarity, Notch signaling, and retinal axon growth.

Dai works on frontier-model evaluation and benchmark design across scientific, technical, and software-engineering domains. His work includes designing and stress-testing research-level evaluation tasks, defining rubrics and acceptance criteria, validating scientific and quantitative reasoning, analyzing model and agent failure modes, and calibrating tasks to distinguish genuine reasoning from shortcuts or superficial success. He also develops software tools and workflows for AI-assisted engineering and scientific computing.