--- a
+++ b/tests/utils/test_io_utils.py
@@ -0,0 +1,53 @@
+import os
+import shutil
+import unittest
+
+import numpy as np
+import pandas as pd
+
+from dosma.utils import io_utils
+
+from .. import util
+
+IO_UTILS_DATA = os.path.join(util.UNITTEST_DATA_PATH, "io_utils")
+
+
+class UtilsTest(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        io_utils.mkdirs(IO_UTILS_DATA)
+
+    @classmethod
+    def tearDownClass(cls):
+        shutil.rmtree(IO_UTILS_DATA)
+
+    def test_h5(self):
+        filepath = os.path.join(IO_UTILS_DATA, "sample.h5")
+        datas = [{"type": np.random.rand(10, 45, 2), "type2": np.random.rand(13, 95, 4)}]
+
+        for data in datas:
+            io_utils.save_h5(filepath, data)
+            data2 = io_utils.load_h5(filepath)
+
+            assert len(list(data.keys())) == len(list(data2.keys()))
+
+            for key in data.keys():
+                assert (data[key] == data2[key]).all()
+
+    def test_pik(self):
+        filepath = os.path.join(IO_UTILS_DATA, "sample.pik")
+        datas = {"type": np.random.rand(10, 45, 2), "type2": np.random.rand(13, 95, 4)}
+
+        io_utils.save_pik(filepath, datas)
+        datas2 = io_utils.load_pik(filepath)
+
+        for data in datas:
+            assert (datas[data] == datas2[data]).all()
+
+    def test_save_tables(self):
+        df = pd.DataFrame({"a": [1, 2, 3, 4], "b": [5, 6, 7, 8]})
+        path = os.path.join(IO_UTILS_DATA, "table.xlsx")
+        io_utils.save_tables(path, [df])
+
+        df2 = pd.read_excel(path, engine="openpyxl")
+        assert np.all(df == df2)