Senior Tools Development Engineer

NVIDIA NVIDIA · Semiconductors · Pune, India

This role focuses on building agentic infrastructure for test automation and quality engineering within the NVIDIA Omniverse platform. The engineer will design and deploy multi-agent systems, orchestration frameworks, and autonomous pipelines, with a strong emphasis on evaluating agent output quality and establishing observability for these workflows. The goal is to enable engineers to ship high-quality software with greater speed and confidence.

What you'd actually do

  1. Develop and deploy multi-agent systems for automated test generation, log analysis, failure triage, and bug-filing workflows
  2. Build and maintain agent orchestration frameworks using tools such as Claude Code, MCP servers, and agent SDK patterns
  3. Create autonomous pipelines that reduce cognitive load on engineers by routing failures, surfacing root causes, and generating actionable bug reports
  4. Build evaluation systems to measure agent output quality — ensuring autonomous pipelines are reliable, not just fast
  5. Establish observability and monitoring for agentic workflows so failures are transparent, debug-gable, and recoverable

Skills

Required

  • Python engineering
  • AI-native development workflows
  • Claude Code
  • Cursor
  • LLM APIs
  • prompt engineering in production
  • building multi-agent or autonomous systems
  • understanding where LLMs fail
  • building mitigations into system design
  • evaluation frameworks for AI-generated outputs
  • test automation
  • CI/CD pipeline design
  • software quality engineering
  • failure analysis
  • test triage at scale
  • reason about test coverage strategically

Nice to have

  • MCP servers
  • agent SDK patterns
  • NVIDIA Omniverse
  • OpenUSD
  • complex platform SDKs
  • custom tool integrations
  • graceful failure recovery
  • retry logic
  • fallback chains
  • human-in-the-loop escalation

What the JD emphasized

  • high-agency engineers
  • autonomous agents
  • agentic infrastructure
  • multi-agent systems
  • agent orchestration frameworks
  • autonomous pipelines
  • evaluation systems
  • agent output quality
  • observability and monitoring for agentic workflows
  • AI-native development workflows
  • multi-agent or autonomous systems that have shipped and run without continuous supervision
  • where LLMs fail — hallucination, context degradation, tool misuse — and experience building mitigations into system design, including evaluation frameworks for AI-generated outputs
  • Built and shipped MCP servers, custom tool integrations, or multi-agent orchestrations that extend LLM capabilities in production
  • Designed evaluation harnesses or scoring systems that measure and enforce LLM output quality at scale
  • built agentic systems with graceful failure recovery

Other signals

  • building agentic infrastructure
  • multi-agent systems
  • autonomous pipelines
  • evaluation systems for agent output quality
  • observability and monitoring for agentic workflows