Current work
Three projects, one theme: AI that stays connected to context.
I’m interested in more than isolated model calls. These projects connect AI to ongoing stories, daily life, and the hardware that runs it.
Marinara · Interactive storytellingA working path from roleplay to animation.
I implemented a MiniMax H3 / ComfyUI flow that turns roleplay narration into a storyboard, grounds characters with their sheets and personas, grounds locations with World Maps artwork, generates keyframes, and animates the resulting shots.
The broader direction is an interactive anime / multiverse experience: combine characters and worlds, switch between roleplay and game modes, make choices, then keep talking with characters in the context of recent adventures. That larger experience is the product direction; the H3 animation pipeline already works.
TypeScriptReactComfyUIMiniMax H3
Manfred · Wearable AIDaily audio, live voice, and visual context.
Manfred combines Omi audio transcription and lifelogging with E09 smart glasses connected to ChatGPT Live. The goal is a wearable interface that can hear daily context, support natural voice conversations, and share what the user is seeing through image uploads.
Voice works today. Photo capture and delivery have been tested and are still being made more reliable.
Demerzel · Local AI operationsKnowledge, briefs, and the machines behind the models.
Demerzel combines daily AI briefs, LLM wiki ingestion from conversations, and management of my local AI machines. Fleet tools monitor supported services, start and stop workloads, and prevent jobs from competing for the same GPU.
Hardware includes an AMD Radeon AI PRO R9700, a dual RTX 3060 system, and a single RTX 3060 system.
Engineering focusBuild the full loop, not just the prompt.
I work across interfaces, model integrations, persistent state, local infrastructure, and automated validation. The interesting problems are often the connections between those layers.
