Principal Software Engineer - LLM Optimization

What you'll do

• Benchmark production LLM workloads, track GPU efficiency, and compare platform performance with external providers.
• Evaluate quantization, speculative decoding, inference engines, and serving architectures; validate changes before production.
• Lead GPU resilience testing and govern agentic AI-enabled engineering workflows with appropriate safeguards.

What they ask for

• Formal training or certification in software engineering concepts.
• Deep hands-on experience with production LLM inference systems such as vLLM, TensorRT-LLM, SGLang, LLM-D, or equivalent.
• Experience with quantization techniques and production tradeoffs; understanding of GPU memory architecture and speculative decoding.
• Experience leading adoption of agentic AI-enabled development practices across teams, including validation, traceability, and secure handling of sensitive data.

Nice to have

• Experience with disaggregated prefill/decode serving architectures and GPU diagnostics such as DCGM/NVML/XID event tracking.
• Experience with ML observability and production monitoring.

Pay

USD 204,250.00-$285,000.00 base salary (Jersey City, NJ); USD 204,250.00-$285,000.00 base salary (New York, NY); USD 204,250.00-$285,000.00 base salary (Seattle, WA); USD 204,250.00-$285,000.00 base salary (Palo Alto, CA)

Tools

vLLM, TensorRT-LLM, SGLang, LLM-D, FP8, INT8/INT4, GPTQ/AWQ, AWS; EKS

AI focus

Optimizes production LLM inference and governs agentic AI-enabled engineering workflows, including AI-assisted code review, test generation, release checks, and incident triage, with validation, security, and governance controls.