Cinematic AI
Cinematic AI · 2026 · Frontend Engineer
Automatic demonstration: open the toolbox, add a sticky note, choose an AI model, configure its prompt and resolution, then reveal two video results on the canvas.
Making AI workflows more flexible — so new models can be added without rebuilding their interfaces.
The problem
The workflow editor initially supported a fixed set of models, each with predefined configuration forms. As the model catalog grew to over 1,000 models, maintaining these forms became increasingly difficult and limited how easily new models could be introduced.
The approach
I built a dynamic configuration system that turns model schemas into UI controls, allowing each model to display its own properties and requirements. I also connected the configuration to validation and workflow execution, so users could configure different models within the same editor.
The outcome
Replaced hardcoded model configurations with a schema-driven system that can adapt to a growing model catalog. This made the configuration experience more flexible and reduced the need to build individual forms for new models.