About Tandemn
Tandemn builds an open-source planning and execution system for AI inference. It coordinates heterogeneous GPU clusters across clouds, regions, VPCs, and on-premises environments as a single pool of capacity. By planning workload configuration and placement together, Tandemn helps teams run existing workloads more efficiently and make room for new ones on the GPUs they already have.
Across evaluations, Tandemn delivers 2.4× effective GPU capacity and up to 92% lower cost per token. Our team brings together researchers, systems engineers, and mathematicians. You’ll work directly with the people building the system and own problems from definition through implementation and evaluation. Learn more about Tandemn.
The role
You’ll build the core systems that let customers deploy and operate Tandemn in their own environments. Ship production code, learn how real infrastructure behaves, and work closely with researchers and engineers to make the system reliable, debuggable, and straightforward to operate. The work spans algorithms and systems. We’re looking for exposure to both, with particular depth in systems engineering.
Responsibilities
Build the systems that make the planner’s decisions work in real infrastructure. Customers may run multiple Kubernetes clusters across clouds and regions; the Tandemn Operator reconciles desired workload states for online and batch inference and tracks resource allocation across those clusters. You’ll extend controllers and reconciliation paths, integrate serving and autoscaling systems such as llm-d and Ray, build request routing across clouds and regions, support heterogeneous GPUs, and develop observability and telemetry. Own how workloads start, reconfigure, recover, and report their state, carrying changes through testing and production operation while feeding serving results back into the planner.
Required qualifications
Minimum two years of software engineering experience, with a strong foundation in systems or infrastructure. Show us a system you built or operated, the decisions you owned, and how you debugged it when something went wrong.
- Strong software engineering fundamentals: readable, tested code, APIs, and debugging. Python or Go is preferred.
- Comfort with Linux, Git, and command-line workflows.
- Familiarity with operating systems: processes, threads, concurrency, locks and semaphores, and containers.
- Networking fundamentals, including LANs, NAT, TCP/IP, network layers, ports, and firewalls.
- Familiarity with cloud and distributed systems, including VPCs, services, SQL and NoSQL data models, serialization, and encoding.
- Exposure to inference engines such as vLLM, SGLang, or NVIDIA Dynamo, and techniques such as tensor and pipeline parallelism.
- Readable code, useful documentation, and a willingness to learn as the inference landscape changes.
Nice to have
- Kubernetes fundamentals, including pods, services, ingress, and Helm charts; experience with controllers or operators is especially useful.
- Experience with observability tools such as Prometheus and Grafana.
- Exposure to scheduling or orchestration systems such as Slurm or SkyPilot, or inference infrastructure such as llm-d and Ray.
Traits we value
Treat AI-generated code as unverified. Use tools such as Claude and GPT to move faster, but review and test their output, understand the architecture and tradeoffs, and own the code you ship.
