Software Engineer, Compute Platform

Supercomputing built for AI.

Cotidal is building the computing infrastructure that AI companies and research labs need to train and run their models. Starting with TPUs, we design the hardware and software together around their workloads, with the goal of making powerful AI chips easier to use at scale.

Our mission is endless compute for humanity. Over time, we want customers to choose across chip stacks without rebuilding their infrastructure.

You’ll define and build Cotidal’s compute platform, from the software that brings accelerator capacity online to the services AI teams use to provision resources and run workloads. Customers build on this platform directly: its APIs, reliability, and cost are what they experience as Cotidal. You’ll make architectural decisions across provisioning, scheduling, and fleet reliability, carrying systems from their first implementation through production scale.

Working directly with customers and teammates, you’ll identify the problems that matter, set technical priorities, and decide how to solve them. We’re looking for engineers with depth in one area and the judgment and curiosity to work across services, operating systems, networking, storage, and hardware.

Your initial focus will reflect your experience and the team’s needs, with room to take on broader ownership as the platform grows.

Compute Platform & Control Plane: Design and build the APIs, resource models, and distributed services that manage capacity and orchestrate workloads. Own allocation, scheduling, and resource lifecycle, including tenant isolation and safe resource reassignment. Make deliberate tradeoffs between utilization, reliability, and the customer experience as demand and fleet size grow.

Cluster Bring-up &

Qualification:

Own bring-up of new capacity, from installed hardware to clusters customers run on. Build the provisioning, configuration, and qualification software across compute, networking, and storage, with repeatable tests for health, performance, and recovery, and define and verify the criteria for releasing each cluster into production. Partner with the HPC engineers who design our clusters: they bring hardware depth to bring-up, and you bring workload requirements back into the next design.

Fleet Lifecycle & Reliability: Define how the fleet is observed, maintained, and recovered. Build health checks, safe upgrade and rollback paths, failure isolation, and automated remediation. Own production diagnosis and incident response for the systems you build, turning failures into software improvements that reduce recovery time, make the fleet cheaper to run, and keep more capacity available for useful work.

• You’ve designed, built, and operated production infrastructure software—services, controllers, or systems that manage real resources. You’ve carried it through failures, growth, and changing requirements.

• You have deep expertise in a part of the stack—distributed control planes, operating systems, networking, storage, or cluster infrastructure—and follow difficult problems across its boundaries. You work from source code, system behavior, and measurements to establish a cause and verify a fix.

• You can reason about distributed state, concurrency, and partial failure. You design systems that recover interrupted operations, reconcile changes safely, and make their behavior understandable to the people operating them.

• You identify consequential problems before someone writes a specification for them. You work directly with users, decide what is worth building, and carry it through to a reliable system. You make your reasoning clear and change your approach when the evidence calls for it.

Experience in one…