Frontend workbench for AutoML Studio. It covers dataset browsing, annotation project management, image/text/multimodal annotation workbenches, DPO preference labeling, task tracking, deployment views, AI pipeline management, and the batch annotation tool workflow in a single React workspace.
- Dataset list and detail pages for image, text, and multimodal datasets
- Annotation project management and full-screen workbenches for detection, classification, segmentation, pose, LLM conversation, MLLM conversation, and DPO preference workflows
- Task list and task detail tracking
- Deployment overview and deployment detail pages
- AI pipeline template, provider, and binding management
- Batch annotation tool management, run list, run detail, result preview, retry, and incremental execution
- Basic platform settings and bilingual UI support
- React 19
- TypeScript 6
- Vite 8
- Ant Design 6
- Tailwind CSS 4
- Zustand
- Axios
- React Router 7
- Node.js
^20.19.0 || >=22.12.0 - pnpm
>=10
Install dependencies:
pnpm installStart the dev server:
pnpm devThe app runs on http://localhost:3000.
By default, Vite proxies /api requests to http://localhost:45678. The frontend base URL is controlled by VITE_API_BASE_URL.
Create a local env file when needed:
cp .env.example .env.localAvailable variables:
VITE_API_BASE_URL: API base path or absolute backend URL. Default:/api
Examples:
VITE_API_BASE_URL=/apiVITE_API_BASE_URL=http://localhost:45678
pnpm dev: start the Vite development serverpnpm build: run TypeScript build checks and create a production bundlepnpm lint: run ESLintpnpm preview: preview the production build locally
src/
api/ HTTP clients and request wrappers
components/ Shared UI components
hooks/ Reusable React hooks
i18n/ Locale initialization and translation resources
layouts/ Page layouts
pages/ Route-level pages
router/ History and route helpers
stores/ Zustand stores
types/ Shared TypeScript types
utils/ Local utility functions
Create a production build:
pnpm buildBuild the container image:
docker build -t automl-workbench-frontend .Run the image locally:
docker run --rm -p 8080:80 automl-workbench-frontendThe container serves the static frontend and SPA routes. In the root docker-compose.yml, /api is proxied to automl-server:45678.