Scarlett Attensil

About

AI engineer working at the frontier of agentic systems, LLM evaluation, and production AI infrastructure. 7+ years working with ML systems across Meta, private equity, and enterprise organizations including Adobe and Atlassian. At LaunchDarkly I developed the strategic framework around the AI control gap. The core thesis: non-deterministic AI systems require an entirely new layer of development infrastructure that traditional software tooling was never designed to handle. That framework became the foundation of LaunchDarkly's product narrative and directly shaped the AI roadmap. I presented it at AWS re:Invent 2025. On the engineering side I design and build the systems that prove the thesis. Multi-agent architectures using LangGraph and MCP, CI/CD pipelines for AI Configs, online evaluation using LLM-as-Judge, production observability for agentic workflows. If I write about it, I've built it. Right now I'm focused on helping engineering teams understand why AI demands fundamentally different development systems, and giving them the tools and frameworks to build them correctly from day one.

Stats