This role pays more than
85%
of similar roles. Most pay
$170,051–$221,250
— the shaded band above.
At the midpoint, this role pays about
$236k
versus about
$196k
for comparable roles.
Based on 240 similar postings.
Employer
About Nvidia
Nvidia is a leading designer of graphics processing units (GPUs) and system-on-chip units, powering gaming, professional visualization, data centers, and artificial intelligence workloads. Industry: Semiconductors & AI Computing
Nvidia currently has
896 open roles
on FindRole.
Listed pay typically runs
$184,000–$287,500
across 876 roles with salary data.
As a Senior Software Engineer, RL Post-Training Frameworks, you will join the RL Frameworks engineering team to develop open-source tools and infrastructure for reinforcement learning post-training. You will architect and build scalable infrastructure for training-inference-rollout loops across GPUs, CPUs, and LPUs while contributing to frameworks like VeRL, Miles, and TorchTitan. Your daily work involves optimizing distributed runtimes such as Ray and Monarch, ensuring fault tolerance, and managing elastic scaling for long-running jobs. You will leverage Python, C/C++, and PyTorch internals to solve complex challenges in RLHF, PPO, GRPO, and DPO. The role addresses the technical hurdles of coordinating actor, critic, and reward models across heterogeneous hardware to enable models to reason through hard problems and act as autonomous agents within complex environments involving tool-use and code execution.
What does a Software Engineer earn in California?
Median $214000 from 775 postings across 63 companies.
Architect and build RL post-training infrastructure that scales from single GPUs to thousands of nodes.
Optimize training-inference-rollout loops across heterogeneous hardware including GPUs, CPUs, and LPUs.
Contribute to and improve the performance of open-source RL frameworks like VeRL, Miles, and TorchTitan.
Implement fault tolerance, elastic scaling, and fast restarts for long-running distributed training jobs.
Develop systems for CPU-driven rollout workloads such as tool-use, code execution, and agentic environments.
Advocate for researcher needs to internal networking, math library, and compiler teams to prioritize RL capabilities.
Integrate high-performance inference engines into RL training loops for accelerated rollouts.
What we're looking for
MS or PhD in Computer Science, Computer Engineering, or a related field (or equivalent experience).
5+ years of professional experience in distributed systems, high-performance computing, deep learning infrastructure, or ML systems engineering.
Strong proficiency in Python and C/C++.
Experience building or contributing to large-scale distributed systems or runtime frameworks at a frontier AI lab, hyperscaler, or major technology company.
Depth in RL for LLM post-training (RLHF, PPO, GRPO, DPO) and its associated distributed execution challenges.
Expertise in PyTorch internals, including distributed training primitives like FSDP, tensor parallelism, and pipeline parallelism.
Knowledge of Kubernetes runtime internals, including container lifecycle, pod scheduling, and GPU allocation.
Proficiency in end-to-end distributed systems design, including service boundaries, data flows, consistency models, and failure recovery.
PyTorch
JAX
TensorFlow
Python
C++
CUDA
Kubernetes
SLURM
Deep Learning
Reinforcement Learning
LLMs
MLOps
HPC
Distributed Training
Sim-to-Real
Natural Language Processing
Graphics