Python Script for macOS OCR & AI Automation
Budget: $10 – $30 USD
Description:
I need a Python developer to build and polish an automation script for macOS (Apple M2, Sonoma 14.6.1).
The script should:
Hotkey Trigger – When I press /, the script should run.
Screen Capture & OCR – Take a screenshot of my current screen and extract:
The question text
The multiple-choice options (A, B, C, D, E)
The screen coordinates of each choice (so the script knows where to click).
AI Integration – Send the question + choices to an AI model (Together AI API, Hugging Face API, or local Ollama model) and get back the predicted correct answer (A–E).
Mouse Click – Automatically move the mouse to the detected coordinates of the correct choice and click it.
Error Handling –
Handle cases where OCR doesn’t detect choices properly.
Ensure correct Retina scaling for Mac (so clicks align with the right spot on screen).
Configurable AI – Make it easy for me to swap between AI providers (Together AI key, Hugging Face key, or local Ollama).
Requirements:
Strong experience in Python automation.
Familiar with OCR libraries (Tesseract, EasyOCR, or PaddleOCR).
Experience with AI APIs (Together AI, Hugging Face, or Ollama).
Familiar with PyAutoGUI or similar libraries for mouse/keyboard automation.
Must ensure Retina display scaling works properly on macOS.
Deliverables:
A polished, documented Python script I can run easily from Terminal.
Setup instructions (dependencies, virtual environment).
Clear README so I can switch between AI backends.
Note: This is for my personal tests that I make. No specific time limit. Can you make the script work with both DOM parsing (Playwright/Selenium) as the first method and OCR with bounding boxes as fallback if the site uses images or PDFs?
Will the script handle macOS Retina scaling correctly so the cursor lands exactly on the right A/B/C/D option every time?
Can you set it so the hotkey is “/” globally (no matter which app/window is active)?
Will the script allow me to configure the AI backend (Together AI, ChatGPT/OpenAI, or DeepSeek) via a simple config file or .env?
For OCR, will you use something reliable like PaddleOCR or Tesseract, and will it also give me the bounding box coordinates?
Can the cursor move automatically to the detected answer without clicking (just hovering over it)?
I need a Python developer to build and polish an automation script for macOS (Apple M2, Sonoma 14.6.1).
The script should:
Hotkey Trigger – When I press /, the script should run.
Screen Capture & OCR – Take a screenshot of my current screen and extract:
The question text
The multiple-choice options (A, B, C, D, E)
The screen coordinates of each choice (so the script knows where to click).
AI Integration – Send the question + choices to an AI model (Together AI API, Hugging Face API, or local Ollama model) and get back the predicted correct answer (A–E).
Mouse Click – Automatically move the mouse to the detected coordinates of the correct choice and click it.
Error Handling –
Handle cases where OCR doesn’t detect choices properly.
Ensure correct Retina scaling for Mac (so clicks align with the right spot on screen).
Configurable AI – Make it easy for me to swap between AI providers (Together AI key, Hugging Face key, or local Ollama).
Requirements:
Strong experience in Python automation.
Familiar with OCR libraries (Tesseract, EasyOCR, or PaddleOCR).
Experience with AI APIs (Together AI, Hugging Face, or Ollama).
Familiar with PyAutoGUI or similar libraries for mouse/keyboard automation.
Must ensure Retina display scaling works properly on macOS.
Deliverables:
A polished, documented Python script I can run easily from Terminal.
Setup instructions (dependencies, virtual environment).
Clear README so I can switch between AI backends.
Note: This is for my personal tests that I make. No specific time limit. Can you make the script work with both DOM parsing (Playwright/Selenium) as the first method and OCR with bounding boxes as fallback if the site uses images or PDFs?
Will the script handle macOS Retina scaling correctly so the cursor lands exactly on the right A/B/C/D option every time?
Can you set it so the hotkey is “/” globally (no matter which app/window is active)?
Will the script allow me to configure the AI backend (Together AI, ChatGPT/OpenAI, or DeepSeek) via a simple config file or .env?
For OCR, will you use something reliable like PaddleOCR or Tesseract, and will it also give me the bounding box coordinates?
Can the cursor move automatically to the detected answer without clicking (just hovering over it)?
Related categories:
Python
Software Architecture
Mac OS
OCR
Git
Selenium
Automation
API Integration
Hugging Face
AI Development