> Markdown version of [/jobs/ext/4174083-product-owner-conversational-ai](https://www.wearedevelopers.com/jobs/ext/4174083-product-owner-conversational-ai). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Product Owner (Conversational AI)

- **Company:** PENNYMAC
- **Location:** Carrollton, TX, United States
- **Experience:** Expert
- **Salary:** $90,000.0 - $150,000.0
- **Contract:** Permanent contract
- **Skills:** A/B Testing, Artificial Intelligence, Amazon Web Services, Data Analysis, Business Logic, Data Files, Chatbots, Large Language Models, Multi-Agent Systems, Virtual Agents
- **Published:** October 31, 2026
- **Apply:** https://www.careerbuilder.com/job-details/product-owner-conversational-ai-carrollton-tx--9a166be4-bdb4-4378-bb5f-d01c980ef885

## About the Role

Artificial Intelligence (AI), Mortgage, Financial Services, Investment Management, Mortgage Lending, Sales, CDL Driver's License, Artificial Intelligence (AI), Artificial Intelligence (AI) Agents, Team Lead/Manager, * 5-7 years of Product Management or Product Ownership experience, ideally within mortgage, fintech, or complex customer servicing environments.

 * Demonstrated experience with AI/ML products, specifically conversational AI, Large Language Models (LLMs), and deep prompt engineering strategies within an enterprise AWS environment.
 * Proven ability to design multi-agent or complex state-machine conversational architectures.
 * Strong analytical background with a track record of building effectiveness dashboards, running A/B tests, and utilizing AI-specific metrics (e.g., hallucination tracking, prompt latency) to drive improvements.
 * Ability to translate complex business logic into precise technical configurations and natural language flows.
 * Exceptional leadership and stakeholder management skills.

## Description

We are seeking a Product Owner to serve as a prompt-development focused leader for Pennymac's conversational AI initiatives. Operating within a Federated Hub-and-Spoke model, you will sit at the orchestration level across work teams to own the company-wide prompt and persona strategy. Partnering closely with the core engineering team (the "Hub") and leading a team of Product Analysts, you will drive the parallel development and deployment of voice AI agents across CDL/Sales, Servicing, and Fulfillment.

The Product Owner will:

 * Lead the overall product vision for conversational AI agents, owning the company-wide persona, brand voice, and overarching prompt strategy.
 * Architect the multi-agent orchestration layer, defining how primary routing agents interpret user intent and seamlessly hand off tasks to specialized, domain-specific sub-agents (e.g., specific agents for Sales vs. Servicing vs. Fulfillment). Establish the logic boundaries, tool access, and context parameters for each sub-agent.
 * Continuously develop, test, and tune complex system prompts, system instructions, and few-shot examples. Optimize these prompts across the AWS tech stack to balance conversational quality, token efficiency, and response latency.
 * Establish robust tracking frameworks to measure prompt and agent efficacy at a granular level. Monitor specialized metrics such as hallucination rates, contextual accuracy, task completion/containment rates, and user drop-off to drive continuous prompt optimization cycles.
 * Serve as the central orchestrator across CDL/Sales, Servicing, and Fulfillment teams to negotiate and refine business requirements, ensuring parallel development without technical bottlenecks.
 * Define the "Golden Dataset" of test scenarios to feed into engineering's automated LLM-as-a-judge evaluation pipelines. Oversee structured A/B testing initiatives and data analysis to iteratively optimize agent responses against baseline configurations.
 * Define product strategy and track ROI by aligning AI agent performance to key business KPIs, including Customer Satisfaction (CSAT) and Average Handle Time (AHT).
 * Partner with the engineering team to integrate prompt designs with the core infrastructure, latency management pipelines, and real-time safety guardrails.