高级软件工程师，代理工程
Senior Software Engineer, Agentic Engineering
加入NVIDIA深度学习框架组的新一代工程团队（Agentic Engineering）。我们构建代理工作流，以自动化代码生成、测试和调优，覆盖NVIDIA的框架、编译器和开发者工具。该团队是支撑这一技术栈工程师的倍增器。这个全新的机会让你在深度学习框架内部一个高自主性的团队中发挥基础性的技术影响力。我们直接与早期采用者团队合作，将复杂需求转化为可复用、可扩展的基础设施，其他团队可以采用。这项工作处于一个真正罕见的交叉点：现代AI应用于工程本身，而这家公司正为人工智能革命提供硬件支持。
你将负责的工作：
我们的首批客户是NVIDIA的早期采用者工程团队。你需要与他们建立深入的共同理解，识别代理工作流能产生最大影响的痛点。随着这些团队将代理系统集成到生产环境，需求会不断演变，因此你需要与他们一起迭代验证点，共同验证或修改你的计划。作为应用机器学习专家，你需要运用技术判断力，区分持久的架构机会与“一时热潮”的炒作。
这项工作涵盖多个领域。你可能会对编译器基础设施进行代理化处理，使自主代理能够进行高维优化，并在真实硬件上进行闭环验证。多代理协调是核心，从LLM原生工具到与LangChain/LangGraph等框架的定制合作，推动自主循环，应用变更、测量结果、逐步推进并重复。我们将这些系统集成到git原生工作流和CI流水线中，使代理能够针对真实的GPU进行构建、测试和迭代。熟悉NVIDIA最新GPU是基本要求，因为工作目标是支持这些GPU的团队。我们参与跨组织协作小组，分享可复用的代理方法论，帮助整个组织采用有效的方法。
我们需要看到：
- 计算机科学、工程或相关领域的硕士学历
- 6年以上相关经验
- 强大的Python开发能力
- 了解GPU或其他高度数据并行系统
- 有使用和维护AI系统的项目或工作经验
- 有在最少指导的情况下交付复杂项目的记录，包括在适当的时候提出挑战或同步进展
- 有构建工具的经验
查看英文原文
Join the new Agentic Engineering team, within the Deep Learning Framework Group, at NVIDIA. We build the agentic workflows that automate code generation, testing, and tuning across NVIDIA's frameworks, compilers, and developer tooling. The team is a force multiplier for the engineers behind that stack. This greenfield opportunity offers foundational technical influence within a high-autonomy team inside Deep Learning Frameworks. We partner directly with early-adopter teams to translate complex requirements into durable, scalable infrastructure that other teams can adopt. The work sits at a genuinely rare intersection: modern AI applied to the craft of engineering itself, inside a company whose hardware powers the AI revolution.
What you'll be doing:
Our initial customers are NVIDIA's early-adopter engineering teams. You will develop a deep, shared understanding with them, identifying the friction points where agentic workflows would have the highest impact. Requirements will evolve as these teams integrate agents into production, so you will iterate with them on proof points to validate or revise your plans together. As an applied ML expert, you will use technical judgment to distinguish durable architectural opportunities from "tech du jour" hype.
The work spans several areas. You might agent-ify compiler infrastructure to enable autonomous agents to make high-dimensional optimizations, with closed-loop validation on real hardware. Multi-agent orchestration is core, anything from LLM-native tooling to custom work with frameworks like LangChain/LangGraph, driving autonomous loops that apply changes, measure results, ratchet forward and repeat. We integrate these systems into git-native workflows and CI pipelines so agents can build, test, and iterate against real GPUs. Familiarity with NVIDIA's latest GPUs comes with the territory, since the work targets the teams that support them. We contribute to cross-org collaborative group sharing reusable agentic methodology, helping the broader organization adopt what works.
What We Need To See:
- MS in Computer Science, Engineering, or equivalent experience
- 6+ years of experience.
- Strong Python development skills
- Working knowledge of GPUs or other highly data-parallel systems
- Demonstrated projects or work experience using and supporting AI systems
- Track record of shipping complex projects with minimal direction, including raising challenges or syncing at the right moments
- Experience building tools or systems shaped by direct partnership with internal customer or user teams
- Examples of leading technical work through changing requirements and revising direction when evidence demands it
Experience in one or more of the following areas:
- Multi-agent orchestration frameworks (e.g., LangChain, LangGraph) or LLM-based workflow automation
- Compiler infrastructure, intermediate representations, or program transformation
- Autonomous search or optimization over high-dimensional parameter spaces
- Hardware-aware performance optimization for deep learning workloads
- Code generation systems or domain-specific languages (DSLs)
Ways to stand out from the crowd:
- Passion for following the evolution of ML hardware and staying up to date on emerging kernel programming techniques
- Experience building evaluation or testing harnesses, especially for ML systems or multi-agent workflows
- Track record of building internal tools or frameworks that force-multiply engineering teams
- Demonstrated ability to thrive in ambiguous, self-directed environments while remaining humble: communicating with clarity, actively listening, and finding ground truth
- An allergic reaction to "solutions in search of problems"
Your base salary will be determined based on your location, experience, and the pay of employees in similar positions. The base salary range is 184,000 USD - 287,500 USD for Level 4, and 224,000 USD - 356,500 USD for Level 5.You will also be eligible for equity and benefits.
Applications for this job will be accepted at least until September 17, 2026.This posting is for an existing vacancy.
NVIDIA uses AI tools in its recruiting processes.
NVIDIA is committed to fostering an inclusive work environment and proud to be an equal opportunity employer. As we highly value diversity in our current and future employees, we do not discriminate (including in our hiring and promotion practices) on the basis of race, religion, color, national origin, gender, gender expression, sexual orientation, age, marital status, veteran status, disability status or any other characteristic protected by law.Originally posted on Himalayas