Senior QA Engineer: Automation and AI Product Quality

About ValueCore ValueCore is a value-selling platform (web application, Google Sheets integration, analytics, PowerPoint/PDF export) used by enterprise sales teams. We are rapidly shipping AI capabilities: generative AI value models, customer-facing chatbots, a call transcript parser, and AI assistant integrations. Our customers use these features live in sales calls. A bug doesn't sit in a backlog; it appears on screen in front of a customer's VP. About the role Full-time QA engineer and the single owner of product quality. This is not a test-executor role working a ticket queue. You decide what gets tested, automate it, run it daily, and publish the results before anyone asks. What you'll work with Core platform: PHP web application on AWS (EC2 + ECS). This is the customer-facing product, currently with no automated test coverage, and your biggest greenfield. Modern services: NestJS API + MCP server, and a NestJS AI service (LangChain/LangGraph with Gemini, OpenAI, and Claude models). Unit tests and an LLM eval harness with CI baselines already exist here; you'll extend the pattern, not invent it. Frontends: React apps, a Next.js + MongoDB customer portal, and a spreadsheet-based editing experience, with no end-to-end automation today. Infrastructure: Nx monorepo, GitLab CI, MySQL, Docker, Datadog. What you'll do Own the regression suite end-to-end. Audit the current suite, close coverage gaps across the platform (data import/export, account provisioning, charts, analytics dashboards, PowerPoint/PDF export, integrations), and keep it current with every release. Run daily AI regression. Test generative AI model building, customer chatbots, the call-transcript parser, and our AI-assistant integrations every day against reference configurations, with output-quality checks (sources cited, numbers correct, currency handling) and results tracked over time. Our transcript parser already has an eval harness with CI baselines; you'll extend that approach to the other AI features. Automate aggressively. Introduce browser-level end-to-end automation (Playwright) across the core web app and React frontends, plus API-level suites, wired into GitLab CI so a smoke run happens on every push. Use AI coding tools (Claude Code or similar) to write and maintain tests fast. Target: cut a full manual regression pass from hours to minutes. Publish results proactively. Daily automated status to the team, a weekly QA summary, per-release sign-off reports, and a known-issues register republished weekly. Nobody should ever have to ask "did QA happen?" Gate releases. A written test plan per release, edge-case and negative-input testing (malformed numbers, empty states, permissions), basic performance checks, and clear notification when testing starts and ends. Guard key accounts. Maintain account-specific test checklists for our largest enterprise customers and verify their critical flows before any customer-facing session. Partner with developers on testability. Every ticket gets acceptance criteria and a test plan before it ships; you define what "done" means with the developer, not after. Cover the whole surface. Core platform, AI features, the customer-facing portal, analytics dashboards, and automated weekly scans of the marketing website (spelling, broken links, layout). What we're looking for 5+ years in QA, including 2+ years owning test automation (Playwright, Cypress, or equivalent, plus API-level testing). Experience testing AI/LLM-powered features. You understand non-deterministic output, eval-style assertions, and quality checks on generated content, not just pass/fail UI tests. Hands-on with AI coding tools (Claude Code, Copilot, or similar) to generate and maintain test suites. We expect one person plus AI to do what used to take a team. Has been the first or only QA somewhere before: built process from scratch, self-directed, doesn't wait to be assigned scope. Excellent written English and a reporting reflex. You publish status without being chased, flag risk early, and escalate slipping dates before they slip. Availability overlapping at least 5 hours with US Eastern business hours, with same-day turnaround on urgent pre-meeting checks. Comfortable across a mixed stack: legacy PHP, modern Node/NestJS services, and React/Next.js frontends, with SQL (MySQL) for test-data setup and result verification. Bonus: performance/load testing (k6/JMeter), GitLab CI pipelines, LLM eval frameworks, experience with Google Sheets-based products or SaaS analytics.