I design and deploy AI-assisted workflow systems that turn unstructured inputs into structured, production-ready outputs, across content, automation, and publishing pipelines.
My recent work includes building a live n8n automation for my brand Scroll & Tell that connects Instagram DMs and comments to a multi-stage AI pipeline via the Meta Graph API. The system classifies incoming messages using a local LLM (TinyLlama), routes them across six intent paths, and hands off to a multi-turn AI story extraction agent (Phi-3 Mini) that conducts full conversational interviews and saves completed stories to Google Sheets automatically. This replaced a previous ManyChat-based approach, enabling flexible, AI-driven conversations instead of fixed keyword-trigger flows.
I also designed a modular, five-stage prompt pipeline that transforms raw Reddit stories into production-ready short-form video packages (script, caption, CTA, hashtags), reducing per-video prep time from ~45 minutes to ~15 minutes.
My background combines mathematics, digital content production, and prompt-based AI systems. I focus on systems that work consistently without constant oversight: defined inputs, enforced constraints, and clear exit conditions at every stage.
Best suited for teams or founders who need someone who can design, build, and operate AI-assisted workflows, not just execute isolated tasks.
