diff --git a/sweepai/core/test_vector_db.py b/sweepai/core/test_vector_db.py
new file mode 100644
index 0000000000..19f9d88aeb
--- /dev/null
+++ b/sweepai/core/test_vector_db.py
@@ -0,0 +1,81 @@
+import pytest
+import numpy as np
+from unittest.mock import patch, MagicMock
+from sweepai.core.vector_db import (
+    embed_text_array,
+    openai_call_embedding,
+    multi_get_query_texts_similarity,
+    cosine_similarity,
+    normalize_l2,
+)
+
+@pytest.fixture
+def mock_openai_embedding():
+    with patch('sweepai.core.vector_db.openai_call_embedding') as mock:
+        mock.return_value = np.array([[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]])
+        yield mock
+
+@pytest.fixture
+def mock_redis_client():
+    with patch('sweepai.core.vector_db.vector_cache') as mock:
+        yield mock
+
+def test_embed_text_array(mock_openai_embedding):
+    texts = ["Hello, world!", "This is a test."]
+    result = embed_text_array(texts)
+    assert isinstance(result, list)
+    assert len(result) == 1
+    assert isinstance(result[0], np.ndarray)
+    assert result[0].shape == (2, 3)
+    mock_openai_embedding.assert_called_once_with(texts)
+
+def test_openai_call_embedding():
+    batch = ["Test text 1", "Test text 2"]
+    result = openai_call_embedding(batch)
+    assert isinstance(result, np.ndarray)
+    assert result.shape[0] == len(batch)
+    assert result.shape[1] == 512  # Assuming we're using the default 512 dimensions
+
+def test_normalize_l2():
+    x = np.array([3, 4])
+    result = normalize_l2(x)
+    assert np.allclose(result, np.array([0.6, 0.8]))
+
+    x_2d = np.array([[3, 4], [6, 8]])
+    result_2d = normalize_l2(x_2d)
+    assert np.allclose(result_2d, np.array([[0.6, 0.8], [0.6, 0.8]]))
+
+def test_embed_text_array_with_cache(mock_redis_client, mock_openai_embedding):
+    texts = ["Cached text", "New text"]
+    cached_embedding = np.array([0.7, 0.8, 0.9])
+    mock_redis_client.get.side_effect = [cached_embedding.tobytes(), None]
+    
+    result = embed_text_array(texts)
+    
+    assert len(result) == 1
+    assert np.allclose(result[0][0], cached_embedding)
+    assert np.allclose(result[0][1], mock_openai_embedding.return_value[0])
+    mock_openai_embedding.assert_called_once_with(["New text"])
+    mock_redis_client.set.assert_called_once()
+
+def test_cosine_similarity():
+    a = np.array([[1, 0]])
+    B = np.array([[1, 0], [0, 1], [-1, 0]])
+    result = cosine_similarity(a, B)
+    expected = np.array([[1, 0, -1]])
+    assert np.allclose(result, expected)
+
+def test_multi_get_query_texts_similarity(mock_openai_embedding):
+    queries = ["Query 1", "Query 2"]
+    documents = ["Doc 1", "Doc 2", "Doc 3"]
+    mock_openai_embedding.side_effect = [
+        np.array([[0.1, 0.2], [0.3, 0.4]]),  # Query embeddings
+        np.array([[0.5, 0.6], [0.7, 0.8], [0.9, 1.0]])  # Document embeddings
+    ]
+    
+    result = multi_get_query_texts_similarity(queries, documents)
+    
+    assert isinstance(result, list)
+    assert len(result) == len(queries)
+    assert len(result[0]) == len(documents)
+    mock_openai_embedding.assert_called()
\ No newline at end of file