Senior AI Engineer and Agentic Platforms - Network Architecture
Salary not listed • Israel, Tel Aviv • Full-time
Posted 1mo ago
Remote Work Policy
On-site
Employment Type
Full-time
Categories
AI Agent Engineer
About the job
NVIDIA is seeking a Senior Software Engineer to build agentic workflows for their network architecture group. This role involves designing and implementing multi-step agent workflows that operate on a hardware network simulation environment. You will engineer the context that grounds these agents in specifications and source code, and optimize their runtime performance. The goal is to build practical infrastructure that brings intelligent agents to life, supporting architects in their daily work.
Responsibilities
- Build agentic workflows (loops, graphs, multi-step pipelines) for hardware network simulation and analysis.
- Engineer the context for agent workflows using simulation models, specifications, design documents, and source code.
- Collaborate with network architects to translate their workflows into agent-based solutions.
- Optimize the runtime performance of simulation tooling, focusing on execution time, compute cost, and latency.
- Define evaluation and regression testing for agent workflows to measure changes.
- Build observability for agent runs to track actions, failures, and reasons.
- Establish guidelines for secure and reliable agent workflows, including data handling and access control.
- Act as a technical resource for integration issues between agents and internal tooling.
Requirements
- B.Sc. or equivalent experience in Computer Science, Computer Engineering, or related field.
- 5+ years of hands-on software engineering experience with production system ownership.
- Expert-level C++ programming skills and strong Python skills.
- Solid understanding of the full stack, including hardware (memory, I/O, networking, accelerators) and performance bottlenecks.
- Practical knowledge of building systems around AI models, including agent loops, tool interfaces, context retrieval, and common failure modes.
- Understanding of inference serving concepts (request lifecycle, batching, caching, throughput/latency/cost tradeoffs).