AI Trainer

About Kastle

Kastle is building the AI operating system for consumer lending. We partner with some of America’s largest mortgage lenders to transform their contact centers and compliance operations with AI voice agents.

Backed by Y Combinator, Commerce Ventures, and executives from Snapdocs, Google, and WePay, we’re redefining how millions of borrowers interact with financial institutions.

This is a rare chance to join at the ground floor of a company already plugged into large-scale customers and real production use cases.

Role Overview

We are looking for a part-time AI Trainer with experience operating phones in the mortgage industry. You will help improve our AI system by providing feedback on conversations, annotating data, and testing for quality. If you have worked as a customer service or collections agent for a mortgage servicer, this role is a great fit to use your knowledge to train AI in real-world scenarios.

Responsibilities

• Annotate and label data from mortgage-related conversations.

• Give feedback on AI responses to make them more accurate and helpful.

• Train the AI system to handle mortgage servicing phone calls better.

• Test the AI for issues, including response quality, customer experience, explanation quality, and realism.

• Work ~10 hours per week on flexible schedules.

Requirements

• 5+ years a customer service agent for a mortgage servicing company.

• Knowledge of the mortgage industry, including escrow, payment questions, loss mitigation calls, QRPC requirements, servicing transfer questions, and customer conversations.

• Strong attention to detail and communication skills.

• Reliable internet and computer for remote work.

Preferred Qualifications

• 3+ years as a Mortgage Customer Service Supervisor.

• Familiarity with giving feedback to customer service agents.

Benefits

• Flexible part-time hours.

• Opportunity to shape cutting-edge AI technology.

• Work from home.

Please mention the word **FRUGAL** and tag RMTYyLjIyMC4yMzQuMjY= when applying to show you read the job post completely (#RMTYyLjIyMC4yMzQuMjY=). This is a beta feature to avoid spam applicants. Companies can search these words to find applicants that read this and see they're human.