Design and ship multi-step, production-grade AI agents that automate complex GTM workflows end-to-end (not single prompts or prototypes)
Translate business requirements into robust agent architectures, including state management, tool use, orchestration, and guardrails
Build multi-agent systems where specialized agents collaborate, hand off work, and maintain context across workflow stages
Integrate agents with Salesforce APIs (read/write), mapping object models (Leads, Contacts, Opportunities, custom objects) and handling authentication securely
Connect agents to structured business data in Snowflake and reason over it reliably at runtime (with a strong semantic layer mindset)
Partner with IT/Security and Data Engineering counterparts to align on infrastructure, access controls, and production readiness
Define and improve reliability patterns: error handling, fallback behaviors, output validation, monitoring, and post-launch iteration
Deliver in an agile cadence with proper review gates across dev → staging → QA → production (no solo deploys)
Production agentic systems experience: You’ve shipped at least one multi-step AI agent to production that real users rely on daily, and can explain architecture, failure modes, and how you improved it post-launch.

Multi-agent orchestration: Hands-on experience building systems where multiple agents collaborate (e.g., LangGraph, CrewAI, AutoGen, or equivalent patterns)
CRM + business systems integration: Deep comfort integrating with Salesforce via APIs, including auth, object models, and real-world GTM workflows
LLM API proficiency: Strong hands-on experience with Anthropic Claude, OpenAI, or equivalent — tool use/function calling, context management, and reliable parsing in production
GTM domain knowledge: Working understanding of how Sales, CS, Marketing, and RevOps operate (deal cycles, renewals, pipeline, customer health, partner workflows, revenue reporting)
Security-aware architecture: Ability to design for RBAC, data isolation, and auditability; familiarity with auth patterns like OAuth/JWT/SSO (Okta experience is a plus)
Python (production quality): Strong Python with async patterns, logging/observability, error handling, and deployment considerations
Workflow automation + API integration: Experience building reliable integrations across external systems; familiarity with tools like Workato/Zapier is a plus (we use Workato as our MCP integration layer)
Data warehouse connectivity: Experience building agents that query Snowflake (or equivalent) and reason over structured data safely and consistently
Requirements translation: You think in systems and workflows, push back on weak requirements, and propose better approaches grounded in production reality.

Operations AI Engineer – Contract

Key skills