Senior System Software Engineer - Embedded AI Inference

Introduction

NVIDIA is synonymous with innovation, boasting trailblazers who are shaping the world with their forward-thinking approaches. This is your chance to be part of a vibrant community that's redefining the technological landscape. Ready to shape the future of automotive technology with NVIDIA? Apply now to be part of a team that's revolutionizing the industry and driving innovation to new heights. Your potential awaits!

We're hiring a Senior Software Engineer to develop production automotive software for AI inference and agent orchestration in C++. Join us on an exhilarating journey, where you'll build out the foundation for next-generation automotive software applications: in-car agentic AI and inference of cutting-edge AI models (LLM, VLM, VLA). You would have the opportunity to shape cutting-edge AI frameworks that enable unprecedented in-car AI experiences and provide a reliable backbone for a new generation of Autonomous Vehicles.

If you're passionate about building robust, high-performance AI systems that run on GPUs in real vehicles, we'd like to hear from you.

What you'll be doing

• Design, implement, and maintain C++ agentic AI and AI inference solutions for embedded production platforms.
• Integrate PyTorch Deep Learning models into C++ pipelines, and deploy them for real-time inference on NVIDIA GPUs.
• Build and extend testable, modular libraries and components, including interfaces to models, sensor drivers, and vehicle control.
• Profile, debug, and optimize C++ and CUDA code to meet strict latency and throughput targets.
• Collaborate closely with ML researchers, systems engineers, and automotive partners to turn prototype algorithms into production-ready implementations.

What we need to see

• 8+ years of professional software engineering experience, ideally in high-performance safety-critical software, automotive, robotics, or real-time systems.
• Master's or PhD degree in Computer Science or Machine Learning.
• Strong modern C++ (C++14/17 or later): templates, RAII, smart pointers, STL, and experience building large codebases.
• Solid Python skills for tooling, training scripts, and glue code between data pipelines and C++ components.
• Hands-on experience building agentic AI frameworks and with LLM / VLM inference. Experience with LLM and VLM inference and related optimization techniques like speculative decoding, LoRA, MoE.
• Experience developing on Linux: build systems (CMake), debugging (gdb, sanitizers), profiling, and git-based workflows in a CI/CD environment.
• Familiarity with GPU programming and optimization, ideally with TensorRT.

Ways to stand out from the crowd

• Experience with agentic AI, specifically agents based on edge-friendly models (2-7B), including context management, reliable tool calling, and MCP, as well as experience with agentic coding.
• Direct experience with the NVIDIA DRIVE AGX platform.
• Knowledge of AI model optimization and deployment: quantization (INT8, FP8, 4-bit).
• Familiarity with high-performance LLM inference frameworks like TensorRT-LLM or ONNX Runtime.
• Understanding of software quality practices for safety-critical systems (code review, unit testing, static analysis; automotive standards knowledge is a plus) as well as open-source contributions or published work in AI, robotics, or GPU computing.

Work on challenging, real-world in-car AI inference problems where your ML and C++ skills directly impact the cabin experience and vehicle's self-driving capabilities. Collaborate with a talented, multidisciplinary team of researchers, engineers, and automotive experts. Solve hard technical problems at the intersection of deep learning, real-time systems, and production software engineering.

We offer

Germany, Munich

Full time

2026-03-20

Posted Yesterday

JR2015026

Contact

If this opportunity aligns with your background and interests, please apply with your resume and a brief description of relevant automotive AI projects (links to GitHub, publications, or technical write-ups are welcome).