Summary
Alex Mallen is an ML researcher and Member of Technical Staff at Redwood Research focused on AI safety, specializing in scalable oversight, reward hacking, interpretability, and eliciting latent knowledge. With eight years of experience spanning academic and industry labs—including EleutherAI, the Allen Institute, and multiple University of Washington research groups—he bridges rigorous empirical work and practical model interventions. His publications and projects cover when to trust parametric vs. retrieval-augmented models, probabilistic time-series forecasting, and novel anomaly-detection methods, reflecting a strong foundation in both NLP and dynamical systems. Based in Berkeley, he combines hands-on PyTorch modeling and evaluation with writing that clarifies risks and control strategies for advanced models. An under-the-radar strength is his track record of outperforming global competitors in forecasting challenges and translating that competitive rigor into alignment research.
8 years of coding experience
4 years of employment as a software developer
Bachelor's degree, Computer Science, Bachelor's degree, Computer Science at University of Washington