AI Software Engineer

Job Description:
Zoox builds and operates a fleet of purpose-built robotaxis. ZBIT is our AI-powered platform that automates triage, system analysis, and root cause analysis (RCA) across Zoox's issue-tracking workflow - scaling how fast the company diagnoses and resolves problems. We are now extending ZBIT into Manufacturing, integrating AI-guided triage and RCA directly into the production line so that test failures are diagnosed, worked around, and root-caused in minutes instead of hours.


In this contract role you will add engineering muscle to the existing ZBIT team, adapting and extending our existing AI triage/RCA features for the Manufacturing use case. You will build LLM-powered RCA agents and full-stack features and integrate manufacturing data into the agents' reasoning - shipping features fast on a live platform. This is a hands-on feature-building role (~85% coding) centered on AI agents/features and their front-end surfacing; it is not an ML-research role, and the async/stateful service and API-platform work for Manufacturing is a separate role on the team. Day to day you will:

• Build and extend LLM-powered RCA and triage agents - adapt existing ZBIT agents and tools to diagnose manufacturing-line failures, generate hypotheses with supporting evidence, and suggest confidence-scored workarounds.
• Ship full-stack features end-to-end - Python (FastAPI) services and the supporting web/UI, integrated with the manufacturing test platform.
• Integrate manufacturing data sources - historical issue data, diagnostics, FMEAs, wirelists, BOM, firmware-config sources, and Slack - into the agents' context.
• Iterate on agent quality - design prompts and tools, evaluate outputs, and improve real-world accuracy with the team's ML and RCA engineers.
• Move fast in an existing codebase - become productive quickly and deliver against the Manufacturing roadmap with tested, maintainable code.

Required:

• Strong Python - independently owns services, REST/FastAPI APIs, async jobs, and data pipelines, with production-quality, testable code.
• Full-stack capability - can extend a modern JS front-end (Vue, React, or similar) to deliver features end-to-end.
• LLM / agent development experience - has built LLM-powered features or agents (prompt and tool design, structured outputs, output evaluation and iteration).
• Proven fast ramp on existing codebases - productive quickly in unfamiliar code.
• API and data-integration fluency - building/consuming REST APIs and wiring in heterogeneous external data sources.

Nice to have:

• Experience with Bazel or another large-monorepo build system.
• Manufacturing, hardware, or test-data domain exposure (diagnostics, FMEAs, BOM, wirelists).
• Retrieval/data skills relevant to RCA (vector search/embeddings, log/ticket analysis, Databricks or similar); observability, CI/CD, cloud (GCP/AWS) familiarity.

Onsite Expectation: Hybrid - 3 days/week onsite at Foster City. Remote considered for candidates in a time zone within 3 hours of Bay Area (PT), provided they can align working hours to Bay Area hours during critical work periods.