DGXAI/driftcall-site
DriftCall — project site
Static project site for DriftCall, the OpenEnv-compliant voice concierge env under mid-episode schema drift. Built with Vite + React + TypeScript, vendored @chenglou/pretext for hero text measurement, single saffron accent on dark editorial brutalism.
What this Space is
Pre-built static HTML/CSS/JS only — no server, no GPU, no model. The page embeds the live demo Space (`DGXAI/driftcall-demo`) in an iframe and links out to the env Space and the trained LoRA.
Sister artefacts
- 🤗 OpenEnv env Space — `DGXAI/driftcall-env`
- 🎙️ Voice demo Space — `DGXAI/driftcall-demo`
- 🧠 Trained LoRA — `DGXAI/gemma-3n-e2b-driftcall-lora`
- 📦 Source — github
Build
The site is regenerated from canonical sources at the repo root via deploy/frontend_space/build.sh:
cd DRIFTCALL
HF_TOKEN=hf_... bash deploy/frontend_space/build.sh --pushThe script runs npm ci && npm run build in frontend/, copies the produced dist/ into deploy/frontend_space/build/, drops this README on top, and hf uploads the result. The upload is just static files — no Docker, no Python, no runtime.
