I need a robust, production-ready automation system that can spin up many browser profiles at once, carry out tightly controlled interactions inside each session, and take its marching orders from a REST-style API. The goal is to treat every profile as an isolated worker that can receive a JSON command, open the required page, fill forms, click through navigation paths, scrape any requested data, and report results back to the API—all while running in parallel without leaking state or cookies. You are free to choose the underlying stack, but I picture something along the lines of Selenium, Playwright, Puppeteer or a comparable headless solution orchestrated in Docker so horizontal scaling is straightforward. Resilience is important: if one profile crashes the others must stay unaffected, and the system should retry gracefully or surface clear error logs through the API. Because this is a highly targeted build, I will sign off the project once the following acceptance criteria are met: • A launcher script creates, monitors, and tears down N independent browser contexts on demand. • The service ingests API commands (HTTP POST with JSON) and maps each to the correct profile. • Supported actions include open URL, fill field, click element, extract DOM snapshot, and return payload. • Complete documentation and a brief walkthrough video showing parallel sessions in action. If you’ve built similar distributed automations, especially with Python, Node.js, or Java test frameworks, I’d love to see a code sample or repo link so we can hit the ground running.