Lambda function correctly invoked locally, but crashing on AWS

Making a local invoke:

!sam local invoke "export-redshift-s3" --event event.json

Returns a 200 response, indicating the lambda function was correctly executed:

{"statusCode": 200, "body": "Data successfully exported to S3"}
Invoking Container created from asdf.dkr.ecr.region1.amazonaws.com/lambda-redshift:latest
Building image.................

Using local image: asdf.dkr.ecr.region1.amazonaws.com/lambda-redshift:rapid-x86_64.

START RequestId: asdf-9410-47a6-abb0-1c9fdb1ed422 Version: $LATEST
END RequestId: asdf-a82a-474b-b2b9-b569e464db71
REPORT RequestId: asdf-a82a-474b-b2b9-b569e464db71	Init Duration: 0.14 ms	Duration: 9436.90 ms	Billed Duration: 9437 ms	Memory Size: 128 MB	Max Memory Used: 128 MB

The lambda function within the container:

from io import StringIO
import os
import boto3
import json
import pandas as pd
import sqlalchemy
from datetime import datetime
import utils
import logging

logging.basicConfig(level=logging.INFO)

def handler(event, context):
    
    ID_PROJECT = event['ID_PROJECT']
    
    logging.info(f"ID_PROJECT: {ID_PROJECT}")
        
    db_credentials = utils.get_secret()
    
    logging.debug("db_credentials got from secret manager")

    conn_string = (
        f"redshift+psycopg2://{db_credentials['user']}:{db_credentials['password']}@"
        f"{db_credentials['endpoint']}:{db_credentials['port']}/{db_credentials['name']}"
    )

    db_credentials = utils.get_secret()
    engine = sqlalchemy.create_engine(conn_string)

    with engine.connect() as conn:

        ID_EXPERIMENT = conn.execute('''
        SELECT
            asdf
        FROM
            asdf
        ''').fetchone()[0] + 1
        
        now = datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S')
        
        insert_query = f"""
            INSERT INTO
                asdf
                    asdf
                VALUES
                    asdf
        """

        conn.execute(insert_query)
        
        query = f'''
        UNLOAD (
            'SELECT
                *
            FROM
                asdf
            WHERE
                asdf
        )

        TO 's3://asdf'
        IAM_ROLE 'arn:aws:iam::asdf:role/lambda-export-redshift-s3'
        ALLOWOVERWRITE
        PARALLEL OFF    
        FORMAT AS PARQUET;
        '''

        conn.execute(query)
    
    return {
        'statusCode': 200,
        'body': 'Data successfully exported to S3'
    }

The Dockerfile:

FROM public.ecr.aws/lambda/python:3.12

# Copy requirements.txt
COPY requirements.txt ${LAMBDA_TASK_ROOT}

# Update setuptools and install specified packages
RUN pip install --upgrade setuptools && \
    pip install -r requirements.txt

# Copy function code
COPY lambda_function.py ${LAMBDA_TASK_ROOT}
ADD utils.py ${LAMBDA_TASK_ROOT}

# Set the CMD to your handler
CMD [ "lambda_function.handler" ]

From the X-Ray traces, the initialization of the container seems to be correct.

I even gave Lambda's role admin access:

As I am using lambda to unload data from Redshift to S3, the role is attached to the Redshift cluster (since it's working locally, this configuration is good to go).

Any guidance on why the lambda function is not working when invoked?

It takes more than 4 minutes, but it's timed out. It shouldn't take more than 15s as proved in the local invocation.

Thanks in advance for your willingness to help out!