Founder & Core Architect — Dravy (AI orchestration and structured workflow execution)
Built an AI orchestration system that converts user intent into structured plans and executable actions using deterministic workflows and DAG-based execution. Focused on structured output frameworks for analytical reports, dashboards, and simulated dry-runs rather than manual data annotation. The work supports model/tool interaction patterns that are typically required for AI training pipelines and evaluation by enforcing schema-driven, verifiable outputs. • Designed intent capture/parsing into structured intent objects for reliable downstream processing. • Implemented deterministic step-by-step Directed Acyclic Graph (DAG) task execution for consistent outputs. • Developed a multi-system tool orchestration and execution engine routing data across APIs and databases. • Created structured output schemas for reports, dashboards, and execution simulations (dry-runs).