Python Developer (YOLO/Roboflow + ADB Automation) for Pokémon TCG Pocket Bot

I’m looking for a skilled Python developer to build a fully automated Pokémon TCG Pocket bot. You will be responsible for:

Computer Vision Integration

Train and integrate a custom YOLO or Roboflow model to detect in-game elements (HP bubbles, card images) in real time.

Ensure the model runs smoothly on BlueStacks (Android emulator) at common resolutions (e.g., 1080×1920, 720×1280).

ADB Automation Layer

Write robust ADB scripts to launch BlueStacks, navigate menus (Home ? VS tab ? Ranked matches), and perform taps/swipes in-game.

Implement basic error handling: detect pop-ups or loading screens via screenshot and recover back to a known state.

Rule-Based Game Logic Engine

Develop a simple rules engine that reads the vision output (HP values, card presence) and makes “smart” turn decisions (attack, retreat, play Trainer card).

Support meta-deck recognition: detect opponent’s deck name/ID and automatically switch to a counter-deck when needed.

Modular, Extensible Architecture

Structure the code so that Vision, Logic, and ADB Automation are separate modules (e.g., vision/, logic/, adb/).

Provide clear interfaces so we can swap out or update the YOLO model, add new decks, or modify logic rules without rewriting everything.

Documentation & Delivery

Supply a README with setup instructions (install Python dependencies, configure BlueStacks, set coordinate maps).

Include example config files: deck lists, click/swipe coordinates per resolution, YOLO label files.

Add inline comments explaining key functions and any configuration steps.

Project Milestones (Example):

Milestone 1 (Vision & Basic ADB MVP):
• Label ~100 screenshots for HP and card detection.
• Train a YOLO/Roboflow model (? 90?curacy on held-out test images).
• Build a Python script that captures a screenshot from BlueStacks, runs inference, and draws bounding boxes.
• Write an ADB script to open Pokémon TCG Pocket, navigate to the Home screen, and take a screenshot.

Milestone 2 (Game Logic Engine):
• Implement a rule-based decision engine (e.g., “If my Active Pokémon HP ? 20, retreat.”).
• Detect opponent deck name/ID and choose a matching counter-deck from a predefined list.

Milestone 3 (Advanced ADB & Error Handling):
• Enhance ADB scripts to start a ranked match, open booster packs, and handle common pop-ups (e.g., network errors, loading screens).
• Ensure reliability on at least two emulator resolutions.

Milestone 4 (Modularization & Final Testing):
• Refactor code into separate vision/, logic/, and adb/ folders with clear interfaces.
• Test end-to-end on two different BlueStacks profiles (1080×1920 and 720×1280).
• Provide full documentation (README, example configs, troubleshooting guide).