Velda
velda.ioServerless GPUs that developers love, no Docker or Kubernetes needed
Ops & Infraserverless-gpuai-infrastructurebatch-jobsdistributed-trainingdeveloper-toolscloud-computemachine-learning
About
Velda is a serverless GPU platform that lets developers run AI training and batch jobs directly from their local environment using a simple command prefix (vrun). It eliminates the need for Docker image builds, Kubernetes manifests, or dependency management by provisioning compute only when workloads are running. It also offers a browser-based VS Code environment with prebuilt ML templates for individual developers and teams.
Problem
Setting up and managing GPU infrastructure with Docker and Kubernetes creates too much overhead and friction for developers running AI workloads.
For
AI/ML developers and data engineers running GPU training and batch inference jobs
How it works
Developers prefix any command with `vrun` to instantly offload it to cloud GPU resources without changing code, building images, or writing manifests.
Business model
freemium
Status
launched