Senior AI Research Engineer - RLHF & Safety Red-Teaming
As a Senior AI Research Engineer at OpenMind AI, I led the design and deployment of large-scale RLHF pipelines for training and evaluating instruction-following LLMs. My work included developing automated red-teaming frameworks to identify safety failures and iterative reward modeling to improve model behavior. I directly contributed to the creation, review, and expansion of labeled datasets used for pre-training and reinforcement learning feedback. • Led and implemented RLHF with both PPO and custom reward modeling techniques • Developed automated frameworks for model safety evaluation via red-teaming • Improved LLM benchmarks with targeted intervention strategies and policy filters • Managed data quality and robustness checks on labeled corpora used for alignment research