diff --git a/.gitignore b/.gitignore
index a205fd53d2..d3c5ae7f08 100644
--- a/.gitignore
+++ b/.gitignore
@@ -19,3 +19,4 @@ node_modules
 package-lock.json
 cdk.out
 .env
+**/*.dylib
diff --git a/validation_testing/main.py b/validation_testing/main.py
index 68c6504e1d..bf03f13670 100644
--- a/validation_testing/main.py
+++ b/validation_testing/main.py
@@ -25,6 +25,16 @@ def assert_required_env_vars_set():
     if not all([os.environ.get(env_var) for env_var in required_env_vars]):
         raise RuntimeError("not all expected environment variables were set!")
 
+def pre_test_infra_check():
+    # Responsible for making sure that all stacks from previous runs are cleaned
+    # This is needed as sometimes we have resources that are stuck around after tests
+    # That makes the stack deletion impossible, such as ENIs. 
+    # This will ensure that everything is cleaned up before bringing up a new stack
+    for connector in TESTABLE_CONNECTORS:
+      shell_command = f'sh run_cleanup_infra.sh {connector}'
+      subprocess.run(shell_command, shell=True, check=True)
+
 if __name__ == '__main__':
     assert_required_env_vars_set()
+    pre_test_infra_check()
     run_all_connector_release_tests() 
diff --git a/validation_testing/run_cleanup_infra.sh b/validation_testing/run_cleanup_infra.sh
new file mode 100644
index 0000000000..6d025a29fe
--- /dev/null
+++ b/validation_testing/run_cleanup_infra.sh
@@ -0,0 +1,75 @@
+CONNECTOR_NAME=$1
+VALIDATION_TESTING_ROOT=$REPOSITORY_ROOT/validation_testing
+
+# go to cdk dir, build/synth/deploy
+cd $(dirname $(find . -name ATHENA_INFRA_SPINUP_ROOT))/app;
+
+cat <<EOF > .env
+DATABASE_PASSWORD=$DATABASE_PASSWORD
+S3_DATA_PATH=$S3_PATH
+REPOSITORY_ROOT=$REPOSITORY_ROOT
+EOF
+
+# cd back to validation root
+cd $VALIDATION_TESTING_ROOT
+
+### There are few objects that need to be cleaned up manually before attempting to destroy the stack:
+# 1- ENIs (Affects Postgresql Stack mostly)
+# 2- Redhisft Spill bucket can't be deleted because its not empty (Redshift)
+# 3- Redshift Parameter Group is required to be cleaned up manually as well
+
+#### STEP 1: 
+# Glue attaches ENIs to the VPC subnet, if there is one, in order to connect to the VPC, and then will issue a delete of them later, but if they aren't finished deleting by the time we invoke cdk destroy,
+# CFN will not know about the resources and not be able to delete the stack. So, we proactively look for and delete ENIs attached to glue ourselves first
+echo "STEP 1: CLEANING UP ANY ENIs"
+aws ec2 describe-subnets --filters "Name=tag:Name,Values=${CONNECTOR_NAME}CdkStack*" | jq ".Subnets[].SubnetId" | tr -d '"' \
+| xargs -n 1 -I {} aws ec2 describe-network-interfaces --filters "Name=subnet-id,Values={}" | jq ".NetworkInterfaces[] | select(.Description | startswith(\"Attached to Glue\")) | .NetworkInterfaceId" | tr -d '"' \
+| xargs -I {} aws ec2 delete-network-interface --network-interface-id {}
+
+
+#### STEP 2: 
+# Cleanup any stack specific S3 bucket:
+
+# List and filter buckets
+echo "STEP 2: CLEANING UP ANY S3 RESDUE BUCKETS"
+buckets=$(aws s3 ls | grep ${CONNECTOR_NAME}cdkstack | awk '{print $3}')
+
+# Loop through each bucket
+for bucket in $buckets; do
+    echo "Emptying bucket $bucket"
+    aws s3 rm s3://$bucket --recursive --quiet
+
+    echo "Deleting bucket $bucket"
+    aws s3 rb s3://$bucket
+done
+
+#### STEP3:
+# Cleanup Redshift Parameter group 
+# #todo: this is results in latest changes of redshift construct
+#        We need to move back to L2 contruct, once its fixed
+echo "STEP 3: CLEANING UP ANY REDSHIFT PARAMETER GROUP"
+if [ "${CONNECTOR_NAME}" = "redshift" ]; then
+    aws redshift  delete-cluster-parameter-group -- delete-cluster-parameter-group caseinsensitiveparametername
+fi
+
+# once that is done, we can delete our CDK stack if it exists.
+cd $(dirname $(find . -name ATHENA_INFRA_SPINUP_ROOT))/app;
+prefix="${CONNECTOR_NAME}CdkStack"
+
+npm install;
+npm run build;
+npm run cdk synth;
+yes | npm run cdk destroy ${CONNECTOR_NAME}CdkStack;
+
+# Check if the stack is clean up -- Otherwise fail so that On-Call can manually clean up the stack.
+stacks=$(aws cloudformation list-stacks --stack-status-filter CREATE_FAILED ROLLBACK_FAILED DELETE_IN_PROGRESS DELETE_FAILED | jq -r --arg prefix "$prefix" '.StackSummaries[] | select(.StackName | startswith($prefix)) | .StackName')
+if [ -z "$stacks" ]; then
+    echo "FINISHED CLEANING UP RESOURCES FOR ${CONNECTOR_NAME}."
+    # cd back to validation root
+    cd $VALIDATION_TESTING_ROOT
+    exit 0
+else
+    echo "FINISHED CLEANING UP RESOURCES FOR ${CONNECTOR_NAME}."
+    echo "PLEASE CLEAN UP RESOURCES FOR THE STACK MANUALLY."
+    exit 1
+fi