About
Rachel Esco: Overview
I think about AI in every dimension: visual, analytical, linguistic, and human. With a career path that has zigzagged across disciplines, my journey—which once looked like a series of unrelated pivots—has converged into an almost-fated blueprint for the way I now approach AI.
I bring a rare cognitive hybridity to the mobilization of AI: one side maps logic chains; the other lives in the outside-the-boxness of human imagination.
My earliest pre-career roots in Visual Arts give me a unique vantage point that conceives complex AI architecture as clear visual landscapes. I spatialize reasoning paths, hierarchies, and dependencies as networks of possibility. This lets me shape AI systems with enough backbone to scale, enough flexibility to evolve, and enough humanity to preserve the organic rhythms of how people think, connect, and communicate.
Training and evaluating AI
While working as a Team Lead in AI Evaluation at Mercor, I coached AI trainers to test LLMs in reasoning, knowledge, instruction compliance, and writing style. I also pioneered the development of our Prompt Playbook, a detailed guide that explains effective prompt engineering, edge cases and examples of strong rubrics for evaluating responses using objective, falsifiable criteria.
As an AI Trainer at Micro1, I exposed the limitations in research skills of an advanced-thinking LLM. I engineered simple yet research-dense questions to determine whether the model could find, connect, and consolidate fragmented information into a single correct answer. I then built logic maps tracing each piece of information back to its source, giving the LLM a cleaner path from evidence to answer.
How intelligence is built
From my first career as a high school educator, I have a deep understanding of how intelligence is built. I can spot the missing links in reasoning that lead to confusion or hallucination. To ensure successful outcomes, I draw on both reverse engineering from AI and backward design from teaching—to deconstruct large systems into micro connections, then rebuild them to achieve a clear end goal.
Language that moves people
Lastly, from the 10+ years of my previous career in brand copywriting, I understand how writing style, tone of voice and syntax impact audience psychology and persuasion. As a result, I can build scalable communication systems that protect brand voice without producing copy-and-paste AI sludge.
In retrospect, every detour along my career path—which once looked unconventional—has throttled me toward this single convergence point: the future of AI.
