Member of Technical Staff — Inference

RadixArk•Palo Alto, CA

3d•Hybrid

About The Position

RadixArk is seeking a Member of Technical Staff — Inference to push the limits of large-scale AI inference. You will work on the core systems that serve frontier models at scale, optimizing performance, latency, throughput, and cost across thousands of GPUs. This role sits at the intersection of systems engineering, ML infrastructure, and performance optimization. Your work will directly shape how state-of-the-art models are deployed and experienced by users worldwide. This is a deeply technical, high-impact role for engineers who enjoy working close to the hardware–software boundary and solving performance-critical problems at scale.

Requirements

5+ years of experience in systems engineering, ML infrastructure, or performance-critical backend systems
Strong expertise in large-scale inference systems for LLMs or generative models
Deep understanding of GPU architecture and performance characteristics
Experience optimizing latency- and throughput-critical production systems
Strong knowledge of distributed systems and networking fundamentals
Proficiency in Python, Rust, C++, or Go for production systems
Experience profiling and optimizing compute-intensive workloads
Strong debugging skills across system layers (model, runtime, kernel, network)

Nice To Haves

Experience with LLM serving stacks (SGLang, vLLM, TensorRT-LLM, etc.)
Open-source contributions in ML or systems infrastructure
Familiarity with CUDA, Triton, or custom kernel optimization
Experience with batching, KV-cache management, and scheduling strategies
Experience running inference at scale (1000+ GPUs)
Background in HPC or high-performance systems

Responsibilities

Design and build large-scale inference systems for frontier AI models
Optimize latency, throughput, and GPU utilization in production inference
Develop and improve model serving architectures and runtimes
Work on batching, scheduling, and memory management strategies
Collaborate with kernel, compiler, and systems teams on performance optimization
Debug performance bottlenecks across the stack
Drive reliability and scalability of inference infrastructure
Build tooling for observability, profiling, and performance analysis
Contribute to long-term inference architecture and strategy