Conversational-AI QA Tester | Stress-test

**Overview** We're looking for a detail-oriented QA tester to evaluate AI-powered customer conversations inside a web application. Your job is to run approximately **245 scripted conversations** across multiple pre-built conversational agents in a staging environment and judge the overall quality of each interaction. You'll evaluate whether responses feel natural, stay on topic, avoid repetition, follow the intended conversation flow, and handle different scenarios consistently. No coding or technical setup is required—you'll simply play the role of the customer in a built-in test chat, follow a provided conversation script, and complete a short evaluation after each conversation. All conversations are automatically logged. **What you'll do** Work through a provided testing matrix consisting of **245 conversation scenarios** across multiple pre-built agents. These include: * Standard conversation flows * Interruptions and topic changes * Mixed questions and objections * Off-topic and adversarial behavior * Qualification and disqualification scenarios * Scheduling, purchase, and post-action workflows For each conversation you'll: * Follow the provided script as the customer * Complete any specified mock actions (such as simulated booking or purchase steps) * Rate six evaluation criteria * Leave brief notes for anything unexpected or unusual Expected commitment is **approximately 8 hours per day for 3 consecutive days.** **Must-have** * Native or fluent English with an excellent ear for natural conversation * Previous QA experience testing chatbots, LLMs, AI assistants, or other conversational AI systems (or equivalent experience evaluating conversational quality) * Exceptional attention to detail across repetitive testing tasks * Able to commit to three consecutive full working days **Nice-to-have** * Conversation design or prompt engineering experience * Sales, customer support, or SDR experience * Familiarity with LLM observability tools such as Langfuse **Logistics** * Fixed 3-day engagement * Staging environment and detailed testing instructions provided * Results submitted in a shared spreadsheet