Case studies.
A small set, each one with the numbers and the system behind them.
Owned marketing across three brands. Rebuilt HubSpot lifecycle and GTM attribution in 90 days. Reported into the CEO.
The leadership case study. Three brands, three funnels, three customer profiles, consolidated into one attribution stack and one weekly leadership review.
An AI automation platform that took 85+ recurring tasks off the team, compliant GenAI content, and the attribution model leadership uses for budget decisions.
AI marketing in a regulated industry: automation at scale, generative content with compliance gates, and closed-loop attribution.
Cut Amazon ACOS from 81% → 38% in 2024 while scaling daily spend 3×. Four channels live today.
A DTC mints brand bleeding on Amazon Ads. Rebuilt the architecture in 90 days, then took the brand to four sales channels.
30+ posts/month per brand, zero manual queueing. Cowork + Claude + MCPs orchestrate the whole pipeline.
End-to-end content pipeline that takes a brand brief and ships finished posts to social. Built at HAZE for multi-brand agency throughput, now deployed at WEG with Hugging Face MCP for asset generation.
A codified TikTok formula that's produced multiple seven-figure-view campaigns across DTC, hospitality, and B2B.
Not a 'hack.' A repeatable framework covering hook structure, edit pacing, hashtag stack, and distribution timing, run for client after client.
An open-source Model Context Protocol server that allows Claude to read and write HubSpot CRM data.
TypeScript, 13 tools, write operations gated by default. Built to demonstrate the architecture I'd build inside any marketing organization.
Three Airbnb listings became one booking page at aqua.mx, with the copy, photos and ratings read straight from the listings rather than retyped.
A property owner asked for a simple landing page. Rather than collect assets by text, I read his three listings and built from the real data.