Responsibilities

• Invent and ship the demos, agents, and reference applications that headline NVIDIA Metropolis launches, GTC keynotes, and flagship customer engagements - agentic Vision AI, multi-agent workflows, video understanding, VLM pipelines, and model fine-tuning.
• Move from whiteboard sketch to customer concepts and prototype in days, using agentic IDEs and AI-native coding tools as your default development environment.
• Engage directly with customers, partner developers, and the global developer ecosystem - understanding their workflows, pressure points, and roadmaps, and translating that signal into the next reference design.
• Develop training toolkits, demo scripts, reference architectures, technical blogs, and whitepapers that empower the global sales force, partner engineers, and the broader developer community.
• Represent NVIDIA at GTC, industry conferences, webinars, and partner events, and author the bylined posts that accompany the demos you ship.
• Partner with product management and product marketing on go-to-market and developer-engagement strategy for powerful Metropolis capabilities.
• Collaborate closely with the architecture, research, libraries, tools, product management, and system software engineering teams at NVIDIA to influence next-generation Metropolis platforms.
• Drive product capabilities and roadmap shifts into Metropolis, with your contributions named and visible across Product, Research, and Engineering - credited product influence is a core deliverable of this role, not a side effect.
• Travel is required for on-site customer visits, conferences, and developer events (30%)

Requirements

• Bachelor’s or Master’s degree in Computer Science, Electrical Engineering, or a related field (or equivalent experience) with relevant internships.
• Strong experience within applied AI spanning VLMs, agentic frameworks, model training and fine-tuning, and modern agent orchestration (e.g., NVIDIA Agent Toolkit, LangGraph, MCP).
• Fluency with agentic IDEs and AI-assisted coding workflows, which you treat as the primary engine for shipping high-fidelity prototypes and production-grade code at speed - anchored by strong coding fundamentals that let you drop below the abstractions when the work demands it.
• A bias toward building: when you don’t know how something works, your first move is to prototype a small version of it. Public demos, GitHub repos, or shipped projects speak louder than slides.
• Outstanding communication and customer-empathy skills - you genuinely enjoy being in front of developers, customers, and conference audiences, and you bring fresh ideas and a point of view to every conversation.
• Comfort with the unknown and a high-agency operating mode: given a one-sentence problem statement, you return with a working artifact and a crisp argument for what to build next.
• Brings original ideas to the team - a brainstorm partner who arrives with a take, not a status update.
• A track record of turning customer signal into pioneering product outcomes - features, models, or roadmap shifts that exist because you advocated for them.

Preferred

• Hands-on experience across NVIDIA’s Physical AI and agentic stack - Metropolis, Nemotron, NVIDIA Agent Toolkit, NIM, AI Blueprints, Cosmos, DeepStream, Isaac, and Omniverse.
• Shipped agentic Vision AI applications in production - multi-agent orchestration, tool use, agent skills, and evals.
• Foundation model post-training experience (SFT, RLHF, distillation) for vision and multi-modal models.
• A public footprint - repos, demos, or talks - that reflects an AI-tool-native development style.
• A demonstrable record of customer focused product influence - features, models, or roadmap shifts traceable to your work.