8+ years of engineering experience, with significant portion shipping ML/LLM features to millions of active users. Proficient in Python/PyTorch. Comfortable with modern LLM stack (vLLM, HuggingFace, fine-tuning pipelines, eval frameworks). Understand the technical rigor required to moderate NSFW content effectively. Intuition for LLM alignment (prompt conditioning, temperature, sampling). Value velocity and ability to distinguish between academic and shippable solutions. Obsess over metrics, regressions, and user experience.