英伟达SAI Verification Engineer
任职要求
• B.Sc. degree or equivalent experience in Engineering/Computer Science/related field.• 4+ years of experience as a Software Engineer.• Intrinsically motivated with a desire for automation programming.• Strong programming skills in Python.• Strong technical abilities, problem solving skills, coding and design skills.• Ability to lead feature development, take full ownership and deliver independently.• Linux knowledge: have a general understanding of Linux operation system concepts.• C experience and extensive knowledge.• Ability to understand, debug and improve 3rd party complex code.• Excellent communication in English and leading skills. Ways to stand out from the crowd: • Knowledge in one or more of the following Networking areas: Ethernet, VLANs, TCP/UDP/IP, QoS, L2-L3 protocols.• Prior software testing experience, with an understanding of Software Testing Tools and Methodologies.• Experi…
工作职责
NVIDIA has been transforming computer graphics, PC gaming, and accelerated computing for more than 25 years. It’s a unique legacy of innovation that’s fueled by great technology—and amazing people. We are looking for a SAI software verification engineer to join our SAI development team. As a Verification Engineer at NVIDIA, you will work closely with NVIDIA’s development and architecture teams responsible for SAI design and gain deep understanding of NVIDIA products and technologies. What you’ll be doing: • Be part of NVIDIA SAI multi-national R&D team, contribute code to SAI community and Nvidia SAI implementation.• Design and implement robust, maintainable, and efficient automation test suite.• Work with continuous integration systems, regression tools, automate builds, run test suites, generate test reports, isolate and classify failures and review new degradation.• Work with experienced teams which are well known in the SAI community.• Develop high quality code, most of the code is open source and published and reviewed in industry leading open source environments.
【核心职责】 1. 3D空间-语言对齐数据集建设与评测 负责构建和维护面向车载场景的“物体3D位置-自然语言”对齐数据集，制定数据标注规范与质量验收标准。 设计并实现多维度评测指标体系（如3D Grounding Accuracy、Spatial Reasoning Score等），量化评估VLM/LLM在车内视角下的3D感知与指代理解能力。 搭建自动化数据闭环流水线，挖掘智驾感知数据中的长尾Corner Case，持续迭代高质量SFT/RLHF训练数据。 2. 座舱Agent框架开发与智驾交互 设计并开发高实时性的座舱Agent软件框架，支持多模态输入（语音、手势、视线、3D感知）与工具调用（车控、导航、智驾指令）。 实现Agent与智能驾驶系统的标准化交互接口，使Agent能够基于3D环境理解主动发起智驾意图确认、风险预警或接管建议。 优化Agent的规划（Planning）与记忆（Memory）模块，提升多轮对话中的上下文保持能力与任务执行成功率。 3. 模型部署与端云精度对齐 主导模型量化（INT8/FP16）、算子融合及内存管理，解决端侧部署中的精度损失问题，确保端侧模型与云端基座模型的输出一致性达到业务验收标准。 建立端云精度对齐的自动化回归测试机制，保障模型版本迭代过程中的稳定性与可靠性。
- 搭建自动驾驶软件模块, 搭建功能仿真和功能开发工具链； - 与硬件、算法、工程集成团队合作，优化自动驾驶软件系统； - 负责特定场景自动驾驶项目的工程实现。
核心职责： 1. 安全开发（基于AI coding工具,比如 claude code) - 负责自动驾驶系统功能安全设计，开发&测试准出；能够基于前沿的AI coding工具，高效建设功能安全主线/项目产品，比如基于claude code 设计相关skills/agents,助力功能安全工作产物快速交付； - 基于ISO26262, ISO/PAS 8800, ISO21448，GB 等标准开展安全生命周期中的各种安全活动。 2. 客户功能安全管理 - 同时担任多客户功能安全经理角色，进行客户产品安全生命周期管理与技术支撑。基于公司功能安全主线产品，提供一整套客户功能安全解决方案。
工作职责： 负责自动驾驶端到端算法工程化方案设计、软件框架实现和落地。 负责自动驾驶端到端算法和工程交付和评测体系的设计和落地。