Principal Product Manager

Job Description

Sr / Principal Product Manager - AI Agent, Orchestration & Evals

Location: Nashville, TN or Remote - USA (Bay area)

About [REDACTED]

The world revolves around technology, and so do we. As a leader in tech care, our core business is providing protection, repair, installation and expert support for all our favorite devices. When tech breaks, we fix it. When it changes, we evolve. It's how [REDACTED] creates meaningful solutions to millions of customers everyday—who choose us because we care.

The work we do is hands-on, collaborative, and most of all: vital. We don't just solve tech problems; we relieve the stress and frustration that comes with them. Our team takes ownership, and they put customers first. We're looking for smart, resourceful and passionate people to join us.

About the Role

We are hiring a Senior or Principal Product Manager to own the brains of Joy.ai — [REDACTED]'s AI-native tech concierge. This role owns the agent itself — the reasoning runtime, how it orchestrates across skills, tools, and other agents, and the evaluation systems that prove it is useful, trustworthy, and ready for real customers.

This is a hands-on product-building role. You will turn ambiguous household problems — a cracked tablet, a stalled activation, a claim, an upgrade across multiple carriers — into prototypes, evals, launch scopes, and shipped agent capabilities, working shoulder-to-shoulder with engineering, design, data science, and domain teams.

You do not need years with the title "AI PM," but you do need to learn new technical domains quickly, build instead of wait, and turn ambiguity into shipped customer value. In your first six months, you will own the agent's quality bar end-to-end: establish the eval harness as the source of truth for shipping decisions, define the agent-to-agent orchestration, and raise the reliability and breadth of Joy's reasoning across our highest-value household journeys.

Responsibilities

• Own the Joy agent reasoning layer end-to-end — the runtime loop that advances the agent through model calls, tool execution, and streamed events — from opportunity discovery and product thesis through launch, adoption, and iteration.
• Shape how Joy interoperates with the broader [REDACTED] stack.
• Own evals as the product's quality system. Define product-quality evals — offline and online — that measure whether the agent is useful, trustworthy, and ready for customer use.
• Partner deeply with engineering on reliable agent architecture: evaluation harnesses, observability and tracing, human-in-the-loop controls, guardrails and production readiness.
• Build conviction through working demos, measurable eval results, customer discovery, and clear decision memos.
• Rally domain PMs, design, engineering, data science, field, legal, security, and executives around the decisions needed to ship an agent that is both breakthrough and dependable.

About You

Basic Qualifications

• 8+ years of product management.
• Track record of 0-to-1 product development: you have taken something from concept to first customers, revenue, or measurable business impact.
• Experience using customer evidence, product metrics, and structured judgment to make high-stakes product decisions with incomplete information.
• High agency with modern AI tools: you use tools such as Claude, Cursor, Codex, or equivalents to prototype, analyze, write, test, research, or automate parts of your own product workflow.

Preferred Qualifications

• Hands-on experience building, prototyping, or shipping AI agents, LLM features, agent orchestration, workflow automation, or decision-support systems.
• Strong intuition for agent evaluation and reliability — eval harness design, offline vs. online evals, hallucination and failure-mode analysis, human-in-the-loop design, permissions, auditability, and tracing/observability (e.g., Braintrust, Sentry, LangSmith, or equivalents).
• Experience managing design partner, Early Adopter, or lighthouse customer programs.
• Strong product taste: you can simplify complex, ambiguous household journeys into opinionated, trustworthy agent behavior that customers understand and adopt.

You'll Be a Great Fit If

• You are already building with AI, whether at work, through side projects, prototypes, internal tools, or hands-on experimentation.
• You have strong opinions about where agents should act autonomously, where humans must stay in control, and how to prove the difference with evals rather than assert it.
• You are candid about tradeoffs, including projects you killed, launches that missed, or bets that changed direction because the evidence demanded it.

Less

More