Senior Software Engineer, Agent Simulation and Evaluation

at Nvidia

USD 152,000-287,500 per year

SENIOR

✅ On-site

Tech Stack
Tag name is followed by "@" symbol and proficiency level value. About proficiency levels:

1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;

3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;

7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;

10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.

AI @ 7 Agentic AI @ 6 Agentic Systems Claude Code @ 4 Codex @ 4 Experimentation @ 6 GPU @ 4 HPC LLM Performance Analysis @ 4 Profiling @ 4 Python @ 7 Software Development @ 4

Details

NVIDIA is looking for a Senior Software Engineer to develop agent simulation and evaluation systems for emerging AI workloads. The role focuses on using agentic methods to streamline simulation, evaluation, performance analysis, and engineering workflows for NVIDIA's LPU organization.

Demand for inference is growing rapidly due to reasoning models, coding agents, parallel subagents, and long-horizon tool use. These workloads increase the complexity of evaluating, simulating, and optimizing AI systems across GPU and LPU platforms. The successful candidate will combine an understanding of model architectures with performance analysis across GPU, LPU, and heterogeneous systems.

Responsibilities

Build agentic systems that automate simulation, evaluation, performance analysis, and reporting.
Develop tools to configure experiments, run evaluations, analyze results, identify regressions, and recommend follow-up work.
Analyze the inference characteristics of LLMs, reasoning models, coding agents, multimodal models, and agentic harnesses.
Build and validate performance models for GPU, LPU, and heterogeneous GPU-LPU systems.
Collaborate with inference, hardware, runtime, compiler, evaluation, and product teams to improve simulation quality and accelerate engineering decisions.
Automate critical LPU workflows, including benchmarking, workload characterization, capacity planning, and release qualification.

Requirements

Strong experience building AI agents or AI-backed engineering systems.
Familiarity with modern model architectures and AI inference workloads.
Experience evaluating AI models or agents and analyzing performance.
Experience with simulation, profiling, benchmarking, or performance modeling.
Strong Python skills and experience building reliable engineering tools.
Master's degree in Computer Science, Engineering, or a related field, or equivalent experience.
Five or more years of relevant software development experience.
Two years of experience building AI agents or AI-backed engineering systems.

Preferred Qualifications

Experience using coding agents such as Codex, Claude Code, or similar tools to automate technical workflows.
Knowledge of GPU or accelerator architecture, distributed inference, or high-performance computing.
Open-source projects demonstrating agentic AI, evaluation, or performance-engineering applications.
A rigorous analytical approach to experimentation, data collection, and validation.

Compensation and Benefits

The base salary depends on location, experience, and the pay of employees in similar positions:

Level 3: $152,000-$241,500 USD per year
Level 4: $184,000-$287,500 USD per year

The role also includes eligibility for equity and benefits. Applications will be accepted at least until August 16, 2026. This posting is for an existing vacancy. NVIDIA uses AI tools in its recruiting processes and is an equal opportunity employer.