refactor load_expert_model to disable quantization for stability and update memory usage estimation

refactor model loading functions for improved readability and maintainability

apply dynamic quantization to expert and translation models for improved CPU performance

add model.pt file for versioning with Git LFS

add libgl1 and libglib2.0-0 to system dependencies in Dockerfile

add opencv-python-headless to requirements.txt

add torchvision and torch to requirements.txt

move libgl1 and libglib2.0-0 to packages.txt

add missing dependencies: libgl1, libglib2.0-0, and opencv-python

push

push all

switch model for reasoning

update this crew pipeline

push prompt chnages

push the weather

Merge remote initial commit with TerraSyncra AI deployment

Initial TerraSyncra AI deployment - CPU optimized with lazy loading and Qwen 1.8B model

initial commit