I've spent 20+ years in Adobe tools and 15+ years making video, so when generative AI showed up I didn't treat it as a novelty. I treated it like a new camera. My work runs in two lanes: enterprise tools like Adobe Firefly and Custom Models, where brand safety and licensing matter, and open-source pipelines like ComfyUI running on my own hardware, where I can push further and learn faster. Most clients need a bit of both.

Pre-Production:

• Figure out where AI actually saves time or adds value in a project, and where a traditional shoot is still the better call.

• Write structured prompts and shot lists that get consistent, usable results instead of one lucky frame out of fifty.

• Build and train custom models (Firefly Custom Models, LoRAs) so generated content stays on brand.

• Storyboard with AI to test concepts fast before anyone spends real money.

Production:

• Generate video and imagery across Adobe Firefly, Google Flow/Veo, Seedance, Krea, and local models.

• Run a local ComfyUI pipeline on an RTX 5080 workstation for private, repeatable, cost-controlled generation.

• Keep a human in the loop for every creative decision. The tools generate options; I pick what works.

Post-Production:

• Upscale and restore footage with SeedVR2 and other ComfyUI workflows.

• Finish in Premiere Pro and After Effects, blending AI shots with live footage so the seams don't show.

• Check output against brand standards before it goes anywhere public.

Beyond Content: AI-Built Tools

I also build small AI-powered apps that solve real workflow problems, including a Brand Consistency Auditor, a brand asset pipeline, and a Notion router using MCP. None of these needed an engineering team. Read more in my post on vibe coding.

Credentials: Adobe Certified Professional in Firefly, Firefly Custom Models & Services Foundations, Adobe Education Leader (6+ years), MIT Sloan Digital Marketing Analytics, MBA.