Principal Engineer, Model Optimizations

📋 Description

• Setting the technical strategy for model optimization across the fleet: which techniques we invest
• Owning quantization end to end—FP8, FP4/MXFP4, INT8, weight-only schemes such as AWQ and
• Driving performance for modern architectures at the execution level: MoE routing and fused expert
• Leading speculative decoding work (draft models, EAGLE/Medusa-class methods, n-gram and lookahead
• Writing and tuning kernels where upstream falls short—CUDA, Triton, CUTLASS, and their ROCm
• Establishing how we choose parallelism layouts (TP, PP, EP, attention DP) per model, per GPU

🎯 Requirements

• 12+ years in performance-critical systems, with substantial recent experience optimizing LLM
• Deep understanding of GPU architecture and the inference performance model: memory bandwidth vs.
• Hands-on kernel-level experience on at least one vendor stack and the demonstrated ability and
• Practical quantization expertise, including the judgment to know when a technique that benchmarks
• Familiarity with the internals of at least one major serving engine (vLLM, SGLang, TensorRT-LLM)
• Strong Python and C++/CUDA skills, and rigor in profiling with Nsight, rocprof, and equivalent

🎁 Benefits

• We innovate with purpose. You’ll be a part of a cutting-edge technology company with an upward
• We prioritize career development. At DO, you’ll do the best work of your career. You will work with
• We care about your well-being. Regardless of your location, we will provide you with a competitive
• We reward our employees. The salary range for this position is based on market data, relevant years
• DigitalOcean is an equal-opportunity employer. We do not discriminate on the basis of race