Tensorlake-Ingest

Provides composable infrastructure for running AI agents, including isolated microVM sandboxes, hosted Git, and a versioned POSIX filesystem to deploy agent harnesses, run RL rollouts and evals, and build scalable CI without sharing the app server runtime. Serves developers, data teams, and operations teams in B2B organizations building and testing agent workloads. Delivered as SaaS and infrastructure APIs with serverless functions, cloud volumes, and an option to run inside the customer's own AWS, GCP, or Azure account.

Key features

• Isolated microVM sandboxes
• Snapshot, pause, fork and resume
• Versioned mountable shared directories
• Serverless functions and workflows
• Cloud dev boxes with SSH and IDEs
• Persistent and ephemeral sandboxes
• Stateful sandboxes with auto-suspend
• POSIX directory mounts
• Hosted Git with snapshots
• Mount repository without cloning
• PTY over WebSocket
• Portable sandbox identity
• VPC peering and private endpoints
• Firecracker microVM isolation