Where durable execution meets real-world compute — and you help us make workers and agent compute effortless.
Temporal Cloud split responsibilities historically: Temporal runs the server; customers run workers. On one hand, serverless-native teams, customers with AI/GPU workloads, and organizations without dedicated platform teams all need compute models beyond always-on, self-managed workers. On the other end of this spectrum, even large enterprises with expert platform teams struggle to manage the lifecycle, capacity planning and resource optimization of Temporal workers.
Our charter is to simplify and manage compute for workers and other aspects across the Temporal Cloud.
Compute owns the roadmap that progressively reduces the customer burden of “running workers” — from bring-your-own serverless integrations to fully managed and hosted compute models.
If you like building systems that sit at the intersection of distributed control planes and developer experience, this is it.
Server-side dispatch to compute providers
Invoking and routing work to serverless and container platforms reliably, with strong guarantees and clear failure modes.
Worker operations as a product
Deployment, rollouts, tuning, versioning, and cost optimization — automated, opinionated, and safe.
Isolation for untrusted / dynamic code
We’re exploring primitives like microVM-based sandboxes and ephemeral workers to run workload-scoped compute safely.
Observability end-to-end
From workflow → activity → inner execution (LLM calls, tool invocations, retries), with cost attribution and debugging that works in production.
| Rhythm | Details |
|---|---|
| What we value | High-leverage infrastructure, pragmatic execution, and product-minded engineering that measurably reduces customer friction. |
| Project shape | Meaty efforts that take weeks/quarters, with iterative milestones and “ship value early” prototypes. |
| Collaboration | We partner closely with Product, Cloud Global Services, OSS Server, DevProd, and teams building agent experiences. |
| Operating rigor | We care about safe rollouts, baked testing, and production-grade observability. “It works in prod” isn’t optional. |