THE DURABLE EXECUTION CONFERENCE FOR AI, BY TEMPORAL

MAY 5-7, 2026

SAN FRANCISCO, CA

_/  SPEAKERS  / 

SPEAKER DISPLAY

Edward Zhu

Edward Zhu

TITLE:Software Engineer

COMPANY:Coinbase

STATUS UPDATE

Edward Zhu is a Software Engineer on the Platform team at Coinbase, where he focuses on infrastructure orchestration and reliability. He has led Temporal self-hosted to cloud migration efforts and is currently driving the adoption of Standalone Activities to replace Coinbase’s legacy background jobs infrastructure. Edward works at the intersection of platform engineering and developer experience, helping Coinbase’s engineering teams move to modern, durable execution patterns at scale.

TALK

Durable Job Processing with Standalone Activities

ABSTRACT

If you’ve ever maintained a background job system in production, you know the routine. Something as simple as sending a welcome email, firing an alert, or delivering a receipt shouldn’t require a team to keep running. But jobs often vanish during deployments. Or, a slow consumer starves everything behind it. Teams might start with a simple queue, bolt on retries, add a result store, write a custom scheduler, build observability from scratch, and end up owning a Tier-0 service nobody wants to maintain.

Standalone Activities are Temporal’s answer to durable job processing. A Temporal Activity can now run on its own — not just as a step inside a Workflow. Same function, same Worker, same deployment. The only difference is how you invoke it. Start with a single durable job, and choose to run it or create a Workflow for multi-step orchestration.

Coinbase is actively migrating production workloads from a homegrown background job system onto Standalone Activities. They’ll share the story behind the move: the use cases they’re running today, the challenges that pushed them off their internal platform — reliability gaps, operational burden, and the hidden cost of a service nobody wants to own — and what it looks like running durable jobs on Temporal in production.

We’ll also dive into how Standalone Activities work under the hood, how they differ from Activities inside Workflows, and what makes them different from traditional job queues.