AI Platform
The AI Platform is a ready-to-deploy environment where you submit your own containerized jobs without managing servers. Each job runs on the tier you choose — a heavily discounted flexible tier for interruptible work, or the production tier — all orchestrated with Kubernetes.
How tiering works
You decide what matters more for each job — cost or certainty. The platform handles the rest.
Interruptible, heavily discounted
Jobs that can pause and resume — batch processing, experimentation, pre-training runs — run on spare capacity at a deep discount. They get scheduled when capacity is available and are safely paused and resumed when it is not.
Top-tier performance, guaranteed priority
For jobs that must finish on time — production inference, client-facing workloads — the production tier reserves capacity and gives your jobs scheduling priority, with stable performance guaranteed.
How it works
No servers to provision and no clusters to manage. Your team keeps building software the way it already does — in containers.
01
Build your workload as a standard container image — the same format your engineers already use every day.
02
Send the job to the platform and choose the tier. No capacity planning, no infrastructure tickets.
03
Kubernetes-based orchestration schedules the job across GPU capacity, autoscales as needed, and recovers it automatically if anything fails.
Billing model
Usage is metered and billed according to the tier you choose. Run experimental jobs on the flexible tier when cost matters most; run production jobs on the production tier when it does. The financing dashboard tracks it all in real time.
Each job is billed against the tier it ran on.
No idle reservations — you pay for the compute your jobs actually use.
Real-time usage and spend tracking, with budget limits and downloadable invoices.
Use cases
One platform that scales from experiments to production without re-platforming.
Long-running jobs on the flexible tier, scheduled around available capacity at a deep discount.
Client-facing workloads on the production tier, with priority scheduling and consistent performance.
Rapid iteration and automated model evaluation — spin jobs up and down without managing infrastructure.
Next steps
The AI Platform is one of Tara Cloud's products. See how it connects to everything else.
Tell us about your workloads, the container images you run, and your performance requirements. A senior member of our team will respond within one business day.