About this role
Zoox is seeking a highly motivated, hands-on Lead AI Engineer to spearhead the end-to-end development and deployment of our Enterprise Gen AI and LLM initiatives, building agentic workflows, integrating enterprise data sources, and enforcing identity-aware security, that transform operations across Procurement, Supply Chain, Legal, Finance, HR and Manufacturing (MES).
In this role, you will:
- Design and deploy production-grade LLM applications and intelligent agents capable of complex reasoning, secure tool usage and autonomous multi-step workflows.
- Architect enterprise data/tool integrations using MCP servers, establishing standardized, reusable connectors across Procurement, Supply Chain, Legal, Finance, HR and Manufacturing (MES).
- Enforce security guardrails via AI gateways - authentication, rate-limiting, logging and policy compliance for all model/agent traffic.
- Implement identity inheritance via SSO, ensuring agents act strictly within the requesting user's permissions & privileges (no privilege escalation, no standing overreach).
- Work with the Information Security team on access control (RBAC/ABAC) for agentic systems and establish observability standards for agent behavior and data access.
- Partner with cross-functional teams to build secure, code-driven AI automation.
Qualifications:
- 8+ years in Data/Software Engineering, with 2+ years deploying GenAI/LLMs in production
- Deep Python proficiency; experience with LangChain, LlamaIndex or provider-native orchestration (e.g., Claude Agent SDK, OpenAI Assistants/Agents SDK or hand-rolled).
- Hands-on experience building/integrating MCP servers or comparable connector frameworks.
- Experienced in AI gateway solutions (e.g. Cloudflare AI Gateway) for governing agent traffic.
- Working knowledge of SSO/identity federation (SAML, OIDC, OAuth 2.0) and propagating identity through agentic tool calls.
- Experience with cloud AI services (AWS Bedrock, Google Vertex AI) and vector databases (MongoDB, Pinecone, Weaviate, Milvus, OpenSearch).