Show HN: Velda – Run jobs with serverless GPUs, without container images

Category: infrastructure

Tags: serverless-gpu, batch-jobs, ai-workloads

Score: 6.3/10 (Innovation: 6, Technical: 6, Documentation: 6, Utility: 7)

Velda is a serverless GPU tool that lets developers run AI and batch jobs directly from their local environment without the need for container images or complex Kubernetes manifests. It creates a seamless bridge between local development and cloud scaling by prefixing commands with 'vrun' or 'vbatch', appealing to ML and data engineers tired of container overhead. The project addresses a real pain point in cloud job execution with a practical, developer-friendly approach.

Target audience: ML engineers, data engineers, devops

Repository: https://velda.io · Go · Apache-2.0 · 79 stars

View on Hacker News