RadixArk
Member of Technical Staff — Training
$200K–$400K
Palo Alto +1 · On-site · Staff · Full-time · Posted Feb 17, 2026
- Location
- Palo Alto
San Francisco - Experience
- 3+ years building or operating distributed machine learning systems, large-scale training infrastructure, or high-performance inference systems.
- Visa sponsorship
- Sponsors employment visas (e.g., H-1B, O-1) for eligible candidates.
About the job
Designs, builds, and operates distributed systems and infrastructure for large-scale LLM and multimodal post-training across GPU clusters.
What you'll do
- Build distributed training, rollout, inference, and orchestration systems for multi-GPU and multi-node workloads.
- Profile and optimize full-stack performance, numerical correctness, reliability, and fault recovery.
- Develop reinforcement learning and agentic post-training infrastructure, and collaborate with researchers to productionize experiments.
What they ask for
- Hands-on experience with post-training systems, training backends, or inference systems for large language models.
- Experience in at least two listed areas, including distributed workload performance, numerical correctness, reliability, post-training infrastructure, multimodal systems, or agent infrastructure.
Nice to have
- Familiarity with RL algorithms such as PPO or GRPO and modern post-training frameworks.
- GPU kernel development, communication-layer optimization, large-scale model training or serving, or key open-source contributions.
Pay
$200,000 to $400,000 USD annual salary (base salary)
Tools
Megatron-LM, FSDP, SGLang, TensorRT-LLM, vLLM, Miles, CUDA, Triton
AI focus
Builds infrastructure for LLM and multimodal post-training, including reinforcement learning, agentic post-training, training, inference, and rollout systems.
Model-written summary · the original posting is authoritative