Skip to main content

Alaya Lab (Workspace)

Alaya Lab (ALab) is a persistent cloud GPU workspace built for AI/ML developers that feels exactly like your local laptop. Forget about buying expensive GPU hardware or wrestling with complex local setups—just open your browser and start coding, training, and running inference from anywhere.

By seamlessly bridging the interactive flexibility of lightweight Cloud Container Instances (CCI) with the sheer compute power of HyperTrain, ALab delivers a completely unified experience.

Use the Lab to write code, debug, train large-scale models, and run inference all in one place. With zero context switching between different tools, ALab radically streamlines your AI workflow, helping you go from a raw idea to a production-ready model faster than ever.

Why build in Alaya Lab?​

  • Your Cloud-Native Laptop: Get the exact look and feel of your local machine, supercharged with cloud GPUs. Access your workspace directly from any browser, anytime, anywhere.
  • Zero Setup, Instant Compute: Skip the hardware costs and tedious environment configurations. Spin up your workspace in seconds and dive straight into development.
  • Seamless Scalability: Start small with interactive coding and debugging via CCI, and effortlessly scale up to massive distributed training workloads with HyperTrain—all without leaving the platform.
  • End-to-End Workflow: Unify your entire AI lifecycle. Develop, test, train, and deploy in a single, frictionless environment designed to maximize R&D efficiency.

Key Features​

  • 1-Click Workspace Provisioning: Skip the tedious infrastructure setup. With a single click, ALab automatically pulls your environment image, provisions compute, and auto-mounts your NAS storage. (Note: Requires active DCU/Cash balance and available quotas).
  • Native VS Code & Jupyter Built-In: Jump straight into out-of-the-box cloud IDEs. Switch seamlessly between VS Code and Jupyter Notebooks with a single click from the side menu.
  • Elastic Compute & Hot-Swappable Configs: Start coding and debugging on a cost-effective CPU (default 2C 4GB). Ready to train? Open the Elastic Config panel to instantly toggle to a high-performance GPU, update your base image (e.g., PyTorch, CUDA 12.8), or inject custom environment variables. The workspace will auto-reboot with your new specs.
  • Smart Hibernation for Cost Control: Never pay for idle compute again. If you close your browser, lock your screen, or put your laptop to sleep for 30 minutes, ALab automatically transitions into Hibernation mode to stop billing. It also auto-hibernates if your DCU balance drops below zero, allowing you to manually power it back on once topped up.
  • Seamless HyperTrain Integration: Go from interactive debugging to massive distributed training without ever leaving the workspace. You can configure, spin up, and track HyperTrain jobs directly from the ALab side panel.
  • Work Like a Local Machine (SSH & Port Mapping): Prefer your own local terminal? Copy the direct SSH credentials from the dashboard to connect your local IDE. Need to test a web app? Easily expose and map custom ports (Ports 22, 8888, and 8889 are reserved for system services).
  • Snapshot to Custom Image: Dialed in the perfect dev environment? While your workspace is running, you can save the exact state as a Custom Image to reuse later or standardize across your team.