Selenium to POST Requests Conversion

I already have a working Selenium-based Python scraper. I now want the exact same workflow rebuilt so that every browser interaction is substituted with direct HTTP POST requests. The goal is to gain speed and run headless on a lightweight server, so please avoid spinning up any browser or WebDriver layer.

A wrinkle: the target site throws a CAPTCHA that is neither reCAPTCHA nor hCaptcha—think of a custom image/logic puzzle instead. I do not yet know which third-party solver or in-house technique will be best, so I’m open to your recommendations as long as it can be automated end-to-end in the new request flow.

What I will hand over:
• My existing Selenium script and notes on each page action.
• Sample credentials and test data for hitting the POST endpoints.

What I need back:
1. A clean Python script (requests or httpx) that reproduces every call through POST, handles cookies/sessions, and reliably bypasses the CAPTCHA.
2. A short README showing how to run it and where to plug in API keys if a solver is used.
3. Clear, commented code so I can tweak selectors or headers later.

If you have prior experience converting Selenium jobs to pure HTTP, especially when a custom CAPTCHA is involved, you’ll be able to wrap this up quickly. Let me know which solver or ML approach you’d lean on, any rate limits I should expect, and an ETA for first tests.