Automation & Agents
Intelligent automation systems and agentic tools. Building software that works autonomously and thinks ahead.
Case study · Posterity
Building an AI Editorial Desk: What I Automated, What I Kept Human, and What Broke
Posterity is a short-form history channel run by a small multi-agent editorial desk. How the desk is structured, where I stay in the loop, and what broke once it was running every day.
Articles of Incorporation Document Verifier (n8n + LLM Extraction)
Built for an internal automation challenge in a regulated financial environment: an event-driven document intake workflow that extracts and validates directors/individuals from incorporation PDFs, compares them to submitted metadata, and produces both an internal summary and a customer-facing message—with full observability via tracing.
Credit Risk Agent: Decision Logs + Reinforcement Learning Policy Loop
A research prototype exploring how structured reasoning traces (“decision logs”) and a lightweight reinforcement learning loop can make credit risk decisioning more inspectable, adaptable, and discussion-ready in regulated settings.