Principal Engineer, Inference Memory and Storage Systems

DigitalOcean seeks a Principal Engineer to define a unified memory and storage substrate for large-scale LLM inference, spanning GPU HBM, host DRAM, local NVMe, and remote object storage. The role involves architecting integrations with vLLM, SGLang, TensorRT-LLM, designing cache-aware routing, owning eviction policies, and driving unit economics. The candidate will set technical direction, mentor senior engineers, and contribute to open source. This is a hybrid role based in Seattle.

15+ years building large-scale distributed systems, high-performance storage, or ML systems infrastructure. Deep fluency in memory hierarchies (GPU HBM, host DRAM, NVMe, remote/object storage). Experience with distributed caching or key-value systems, low latency under high concurrency. Hands-on with RDMA/NVMe-oF/NVLink technologies. Strong systems programming in C/C++, Go, Rust, or Python. Rigor in profiling across CPU, GPU, memory, network.