Tea Larson Hetrick

Human-in-the-loop expert specializing in fine-tuning workflows, large language model evaluation, and prompt-level optimization for generative AI systems. My focus is on delivering high-signal, structured feedback across RLHF, SFT, HILF, and model ranking environments, ensuring LLMs align with human expectations, ethical frameworks, and applied reasoning standards.

I’ve contributed to high-accuracy annotation and evaluation projects through platforms like Outlier.ai, where my work supports instruction tuning, behavioral QA, and context-aware prompt refinement. My background in cognitive science, neuroscience, and behavioral analysis (with advanced certifications from Yale, Duke, and Harvard) informs my ability to evaluate model outputs not just for logic, but for coherence, safety, and real-world applicability. My skillset is verified at a College, Graduate, or PhD level expertise in physics, biology, business, biology, arts and humanities, math HTML, Coding, STEM, and chemistry.

Whether evaluating generations, benchmarking fine-tuned variants, or ranking completions for relevance and tone, I bring deep reasoning, precision, and contextual awareness to every task. I work best in environments where thoughtful feedback and rapid iteration drive model improvements and research outcomes.

I’m remote-first, highly structured, and able to scale with evolving model architectures. If you're building alignment-focused systems or training LLMs for specific use cases, I’m ready to contribute high-quality feedback at scale—accurately, reliably, and with domain fluency.