Principal Product Manager, Agentic Orchestration

ZoomInfo is where careers accelerate, and they are seeking a Principal Product Manager for their Agentic Orchestration platform. This role involves owning the agentic tools ecosystem, architecting integration strategies, and driving execution excellence to enhance ZoomInfo's multi-agent orchestration capabilities.

Responsibilities:

• Define, build, and evolve the tool ecosystem that powers ZoomInfo's multi-agent orchestration system
• Own the registry of capabilities agents can invoke, the standards for how tools are built, and the governance for how they're deployed
• Own the integration roadmap that expands agent capabilities through MCP servers, API connections, and partner ecosystems
• Determine which integrations unlock disproportionate value, negotiate technical partnerships, and define the abstraction layers that make integrations composable rather than brittle
• Partner with ML engineers on tool selection optimization, prompt engineering for reliable tool invocation, and error handling patterns that make agents resilient
• Own the feedback loops that surface tool failures, identify capability gaps, and translate production signals into platform improvements
• Build the developer experience—documentation, templates, testing frameworks, deployment pipelines—that lets teams ship new tools without deep platform expertise
• Measure success through tool development velocity: how quickly can a new capability go from idea to production agent invocation?
• Continuously evaluate innovations—function calling improvements, multi-tool planning, tool composition patterns—determining which advances warrant immediate adoption versus patient observation
• Translate between worlds: ML engineers optimizing agent reasoning, platform engineers building scalable execution infrastructure, and product teams shipping customer-facing automation
• Establish prioritization frameworks that balance 'keep the lights on' reliability work against strategic capability expansion

Requirements:

• 4-6 years of product management experience with 2+ years focused on AI/ML infrastructure, developer platforms, or integration products
• Direct experience with production agentic systems—you've shipped products where LLMs invoke tools, not just generate text
• Hands-on familiarity with agentic frameworks (LangChain, LangGraph, CrewAI, or similar) and emerging standards (MCP, function calling patterns)
• Track record building platform products where engineering teams are your customers
• Experience with integration architecture—APIs, webhooks, authentication patterns, rate limiting, error handling at scale
• History of defining product direction in fast-moving technical domains where you're often months ahead of established best practices
• You've moved beyond AI hype into practical application. You understand why agents fail—context window limits, hallucinated tool calls, poor error recovery—and you've built systems that work anyway
• You can discuss prompt engineering trade-offs, evaluate tool description effectiveness, and debug agent execution traces
• Python proficiency lets you prototype, validate assumptions, and speak credibly with engineers
• Tools are only as good as the context agents have when invoking them. You understand how to design tool interfaces that give agents what they need to succeed—clear descriptions, appropriate parameters, useful error messages
• You think about context budgets, information density, and the cognitive load you're placing on the orchestrating model
• You've built products that connect systems. You understand the unglamorous realities: authentication flows, rate limits, webhook reliability, schema evolution, partial failures
• You know that 'integrate with Salesforce' is a six-month roadmap item, not a two-week task
• You design abstractions that hide complexity without sacrificing capability
• Your customers are other engineers and product teams. You measure success through their velocity—how quickly they can ship new agent capabilities using your platform
• You understand platform economics: each team adopting your tools increases their value through shared learnings, amortized maintenance, and network effects
• You resist one-off solutions that don't generalize
• You read agent papers, follow framework releases, and prototype new patterns before they're mainstream
• You maintain strong opinions weakly held—your architectural assumptions update as evidence accumulates
• The agentic landscape moves too fast for static expertise; continuous learning is non-negotiable
• You translate between technical depth and business impact fluently
• You can explain to executives why adopting a new tool standard takes three months but prevents six months of technical debt
• You can communicate to engineers why business constraints require shipping 'works reliably' before 'works optimally.'
• You influence through clarity, not authority