Personally develop or review architecture diagrams, product specifications, data contracts, model and agent interaction patterns, evaluation designs, and release criteria — working directly with SQL, data models, transformation frameworks, orchestration, cloud data platforms, AI model APIs, prompts, and agent workflows when critical decisions require first-hand validation, and building reference implementations before scaling them through internal teams and vendors. Architect and build a reusable orchestration harness that binds workflows to approved skills, bounded agents, registered tools, output contracts, approval gates, stop conditions, and auditable outcomes — permitting agent autonomy only within explicitly defined scope and policy, with governed registries and reusable libraries for workflows, skills, agents, tools, context, memory, and models that separate immutable evidence from curated knowledge.