Senior Principal Software Engineer (LLM Inference Optimization)

Location: Remote – United States

Salary: $230,000–$280,000 base plus annual bonus opportunity and potential equity.

Company

hirify.global is a specialist recruitment partner representing an established global AI technology company building real-time conversational experiences for the automotive industry.

What you will do

• Optimize production LLM inference for latency, tokens per second, throughput, GPU utilization, stability, and cost.
• Develop and tune CUDA kernels, including attention and other performance-critical operations.
• Extend inference engines such as vLLM and other high-performance runtimes through low-level code improvements.
• Improve GPU memory efficiency through KV cache optimization, memory layouts, paging, and cache-aware execution.
• Optimize prefill and decode performance for large-context, latency-sensitive workloads.
• Evaluate and deploy quantization, continuous batching, speculative decoding, distributed inference, and other production improvements.

Requirements

• Significant hands-on experience with production ML or LLM inference systems and direct performance optimization responsibility.
• Strong C++ experience, ideally five or more years, with the ability to modify and optimize complex low-level code.
• Hands-on CUDA development, GPU profiling, and kernel-level optimization experience.
• Deep understanding of GPU architecture, memory bandwidth, memory hierarchies, concurrency, caching, compute utilization, and inference bottlenecks.
• Experience with vLLM, TensorRT-LLM, SGLang, or comparable inference runtimes, ideally including modifications beyond standard configuration.
• A record of diagnosing bottlenecks and delivering measurable improvements in latency, throughput, or infrastructure efficiency.
• Direct production deployment or optimization experience beyond proof-of-concept or research environments.
• Ability to explain personal technical ownership, engineering decisions, and measurable results; strong independent judgment and hands-on problem-solving are required.
• Ability to work effectively in a fast-moving, globally distributed team.

Nice to have

• Rust exposure.
• Experience with llama.cpp, QAIRT, SGLang, or other emerging inference runtimes.

Culture & Benefits

• Permanent remote position for candidates in the United States.
• Work on technically challenging LLM inference problems with direct impact on real-time AI products and infrastructure economics.
• Join a commercially proven AI business investing in advanced inference capabilities while retaining meaningful technical ownership.
• Annual bonus opportunity and potential equity in addition to base compensation.