Infrastructure / local control plane

MODDAL makes GPU deployment readable.

A control surface for launching quantized models on Modal GPUs, watching readiness, managing model presets, and keeping the operational detail close to the person using it.

Built withPython · FastAPI · Modal · vLLMInterfaceHTML · CSS · JavaScript

MODDAL / FIELD MAP01 — 03

System intention

Turn a cloud GPU into a legible workflow.Launch, wait, test, inspect. The interface follows the actual sequence instead of hiding it behind a single button.

PyPythonFaFastAPIModalvLvLLM

Applied AI / cloud GPUsLocal build

How the system reads

One control plane, three moments of attention.

The screens move from choosing a model, to testing the live endpoint, to reading the system when something needs a closer look.

Screen record / 01

Launch first. Then make readiness visible.

Model, GPU, context, and the four-stage readiness pipeline sit in one operational frame.

MODDAL dashboard and model launch screen — 01Dashboard & launchSelect the checkpoint, hardware target, and context window before the cloud work begins.

MODDAL vision playground screen — 02Vision playgroundA focused test surface for image understanding and live model output.

MODDAL system logs screen — 03System logsOperational detail stays close enough to explain what the interface is reporting.

Local project / 01

See the next system.

The project pages are built as an index of how the tools actually work, not as a row of decorative screenshots.

Next: Rapido ↗

YAHYA / EL AMRANIBack to selected work ↗