Inference Infrastructure Architect

📋 Description

• Operate and expand the fleet efficiently, maximizing useful inference throughput per GPU-dollar
• Make the product run on it, including serverless inference for the open-weight catalog, and
• Build serverless serving pools, the fleet layer, the platform under it, weight logistics and

🎯 Requirements

• Owned production LLM serving under meaningful traffic and latency constraints; scale on the order
• Kubernetes on GPU fleets, operated end to end: GPU Operator, device plugins, node pools and
• Deep operational command of vLLM or SGLang: deploying, tuning and upgrading it (parallelism with TP
• Performance engineering at the system level: you read engine and DCGM metrics, reason from
• Python and Go for automation; real Linux, networking and storage depth.
• Work with the community in English and Chinese, and write runbooks and design docs people actually

🎁 Benefits

• The newest hardware, ours from the metal up, with a globally expanding B300 fleet that Telnyx
• Own the platform, build the team, with a greenfield inference platform where you set the pattern
• Open-source first, with upstream contribution as part of the job and conference travel supported.
• Based in mainland China, no relocation required, working remotely with a global, async-friendly