GEOS-CF ML downscaled forecasts

Development

1. Install uv if not installed.

2. Locally, sync and lock the python environment.

   uv sync
   uv lock

3. Confirm that everything works locally

   uv run python -c "from MLpred import mlpred, funcs"

4. Build Docker image:

   docker build -t mlpred:latest .

5. Test Docker image:

   docker run -it --rm mlpred:latest \
       uv run python3 -c \
       "from MLpred import mlpred, funcs"

6. Configure AWS:

   # Also set AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and AWS_SESSION_TOKEN
   export AWS_REGION="us-west-2"
   export ECR_REPO="466270585360.dkr.ecr.us-west-2.amazonaws.com/geos-cf/mlpred"
   
   aws ecr get-login-password --region $AWS_REGION | \
       docker login --username AWS --password-stdin $ECR_REPO

7. Tag and push Docker images

   docker image tag mlpred:latest "$ECR_REPO:latest"
   docker push "$ECR_REPO:latest"