Nuro
Software Engineer, Applied AI Infrastructure
About this role
Nuro seeks a Software Engineer to build infrastructure for trustworthy autonomous AI agents operating within the company's engineering organization. You'll develop closed-loop evaluation systems, agent runtime platforms, and autoresearch automation that apply the same rigor Nuro uses for self-driving vehicles to internal AI tooling.
What you'll do
- Build closed-loop measurement and evaluation systems that track agent acceptance, reversion, and override rates across workflows
- Design and implement agent runtime infrastructure including orchestration, sandboxing, isolation, and permissioning for production systems
- Develop autoresearch automation that enables agents to form hypotheses, run experiments, and evaluate results against training pipelines
- Create observability and auditability layers for agent actions on production code and infrastructure
- Build agent-powered tooling across the engineering lifecycle including code generation, review, and CI/CD automation
- Contribute to post-training of internal models where workloads justify the investment
What they're looking for
- Python or similar systems programming language
- Distributed systems and orchestration
- Data infrastructure and metrics collection
- Machine learning operations and experiment management
- Software security and access control
- Production systems design and reliability
- Observability and monitoring systems
- Agent and LLM frameworks
Opens the official application on the employer’s site. No login required.
Nuro
Nuro builds autonomous vehicle platforms and fleet operations systems, with a focus on reliability, safety, and over-the-air update infrastructure. The company is hiring reliability engineers, software engineers, and operations specialists to improve vehicle hardware resilience, enhance system automation, and ensure fleet operational excellence.
View all jobs at NuroLikely interview questions
- Describe your experience building measurement and evaluation systems for complex automated processes—what made them trustworthy?
- How would you design isolation and permissioning for AI agents acting on production repositories and infrastructure?