diff --git a/.kokoro/continuous/doctest.cfg b/.kokoro/continuous/doctest.cfg
index dca21d43fd..6016700408 100644
--- a/.kokoro/continuous/doctest.cfg
+++ b/.kokoro/continuous/doctest.cfg
@@ -3,7 +3,7 @@
 # Only run this nox session.
 env_vars: {
     key: "NOX_SESSION"
-    value: "doctest"
+    value: "doctest cleanup"
 }
 
 env_vars: {
diff --git a/.kokoro/presubmit/doctest.cfg b/.kokoro/presubmit/doctest.cfg
index dca21d43fd..6016700408 100644
--- a/.kokoro/presubmit/doctest.cfg
+++ b/.kokoro/presubmit/doctest.cfg
@@ -3,7 +3,7 @@
 # Only run this nox session.
 env_vars: {
     key: "NOX_SESSION"
-    value: "doctest"
+    value: "doctest cleanup"
 }
 
 env_vars: {
diff --git a/bigframes/session/__init__.py b/bigframes/session/__init__.py
index 0d7a90c250..072bcc5781 100644
--- a/bigframes/session/__init__.py
+++ b/bigframes/session/__init__.py
@@ -274,6 +274,10 @@ def __init__(
             metrics=self._metrics,
         )
 
+    def __del__(self):
+        """Automatic cleanup of internal resources"""
+        self.close()
+
     @property
     def bqclient(self):
         return self._clients_provider.bqclient
diff --git a/noxfile.py b/noxfile.py
index 92f8acad7f..ef4bf1a37a 100644
--- a/noxfile.py
+++ b/noxfile.py
@@ -697,8 +697,8 @@ def system_prerelease(session: nox.sessions.Session):
 
 @nox.session(python=SYSTEM_TEST_PYTHON_VERSIONS)
 def notebook(session: nox.Session):
-    GOOGLE_CLOUD_PROJECT = os.getenv("GOOGLE_CLOUD_PROJECT")
-    if not GOOGLE_CLOUD_PROJECT:
+    google_cloud_project = os.getenv("GOOGLE_CLOUD_PROJECT")
+    if not google_cloud_project:
         session.error(
             "Set GOOGLE_CLOUD_PROJECT environment variable to run notebook session."
         )
@@ -937,3 +937,31 @@ def release_dry_run(session):
     ):
         env["PROJECT_ROOT"] = "."
     session.run(".kokoro/release-nightly.sh", "--dry-run", env=env)
+
+
+@nox.session(python=DEFAULT_PYTHON_VERSION)
+def cleanup(session):
+    """Clean up stale and/or temporary resources in the test project."""
+    google_cloud_project = os.getenv("GOOGLE_CLOUD_PROJECT")
+    if not google_cloud_project:
+        session.error(
+            "Set GOOGLE_CLOUD_PROJECT environment variable to run notebook session."
+        )
+
+    # Cleanup a few stale (more than 12 hours old) temporary cloud run
+    # functions created by bigframems. This will help keeping the test GCP
+    # project within the "Number of functions" quota
+    # https://cloud.google.com/functions/quotas#resource_limits
+    recency_cutoff_hours = 12
+    cleanup_count_per_location = 10
+
+    session.install("-e", ".")
+
+    session.run(
+        "python",
+        "scripts/manage_cloud_functions.py",
+        f"--project-id={google_cloud_project}",
+        f"--recency-cutoff={recency_cutoff_hours}",
+        "cleanup",
+        f"--number={cleanup_count_per_location}",
+    )
diff --git a/scripts/manage_cloud_functions.py b/scripts/manage_cloud_functions.py
index 6b69089089..33af8463c9 100644
--- a/scripts/manage_cloud_functions.py
+++ b/scripts/manage_cloud_functions.py
@@ -13,7 +13,7 @@
 # limitations under the License.
 
 import argparse
-from datetime import datetime
+import datetime as dt
 import sys
 import time
 
@@ -94,8 +94,10 @@ def summarize_gcfs(args):
         # Count how many GCFs are newer than a day
         recent = 0
         for f in functions:
-            age = datetime.now() - datetime.fromtimestamp(f.update_time.timestamp())
-            if age.days <= 0:
+            age = dt.datetime.now() - dt.datetime.fromtimestamp(
+                f.update_time.timestamp()
+            )
+            if age.total_seconds() < args.recency_cutoff:
                 recent += 1
 
         region_counts[region] = (functions_count, recent)
@@ -106,7 +108,7 @@ def summarize_gcfs(args):
         region = item[0]
         count, recent = item[1]
         print(
-            "{}: Total={}, Recent={}, OlderThanADay={}".format(
+            "{}: Total={}, Recent={}, Older={}".format(
                 region, count, recent, count - recent
             )
         )
@@ -120,8 +122,10 @@ def cleanup_gcfs(args):
         functions = get_bigframes_functions(args.project_id, region)
         count = 0
         for f in functions:
-            age = datetime.now() - datetime.fromtimestamp(f.update_time.timestamp())
-            if age.days > 0:
+            age = dt.datetime.now() - dt.datetime.fromtimestamp(
+                f.update_time.timestamp()
+            )
+            if age.total_seconds() >= args.recency_cutoff:
                 try:
                     count += 1
                     GCF_CLIENT.delete_function(name=f.name)
@@ -134,12 +138,15 @@ def cleanup_gcfs(args):
                     # that for this clean-up, i.e. 6 mutations per minute. So wait for
                     # 60/6 = 10 seconds
                     time.sleep(10)
+                except google.api_core.exceptions.NotFound:
+                    # Most likely the function was deleted otherwise
+                    pass
                 except google.api_core.exceptions.ResourceExhausted:
                     # Stop deleting in this region for now
                     print(
-                        f"Cannot delete any more functions in region {region} due to quota exhaustion. Please try again later."
+                        f"Failed to delete function in region {region} due to quota exhaustion. Pausing for 2 minutes."
                     )
-                    break
+                    time.sleep(120)
 
 
 def list_str(values):
@@ -168,6 +175,19 @@ def list_str(values):
         help="Cloud functions region(s). If multiple regions, Specify comma separated (e.g. region1,region2)",
     )
 
+    def hours_to_timedelta(hrs):
+        return dt.timedelta(hours=int(hrs)).total_seconds()
+
+    parser.add_argument(
+        "-c",
+        "--recency-cutoff",
+        type=hours_to_timedelta,
+        required=False,
+        default=hours_to_timedelta("24"),
+        action="store",
+        help="Number of hours, cloud functions older than which should be considered stale (worthy of cleanup).",
+    )
+
     subparsers = parser.add_subparsers(title="subcommands", required=True)
     parser_summary = subparsers.add_parser(
         "summary",