diff --git a/.github/workflows/test_research_assistant.yml b/.github/workflows/test_research_assistant.yml
new file mode 100644
index 00000000..1d97e92f
--- /dev/null
+++ b/.github/workflows/test_research_assistant.yml
@@ -0,0 +1,48 @@
+name: Unit Tests - Research Assistant
+
+on:
+  push:
+    branches: [main, dev]
+    paths:
+      - 'ResearchAssistant/**'
+  pull_request:
+    branches: [main, dev]
+    types:
+      - opened
+      - ready_for_review
+      - reopened
+      - synchronize
+    paths:
+      - 'ResearchAssistant/**'
+
+jobs:
+  test_research_assistant:
+    name: Research Assistant Tests
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install Backend Dependencies
+        run: |
+          cd ResearchAssistant/App
+          python -m pip install --upgrade pip
+          python -m pip install -r requirements.txt
+          python -m pip install coverage pytest pytest-cov pytest-asyncio
+
+      - name: Run Backend Tests with Coverage
+        run: |
+          cd ResearchAssistant/App
+          python -m pytest -vv --cov=. --cov-report=xml --cov-report=html --cov-report=term-missing --cov-fail-under=80 --junitxml=coverage-junit.xml
+      - uses: actions/upload-artifact@v4
+        with:
+          name: research-assistant-coverage
+          path: |
+            ResearchAssistant/App/coverage.xml
+            ResearchAssistant/App/coverage-junit.xml
+            ResearchAssistant/App/htmlcov/
diff --git a/ResearchAssistant/App/test_app.py b/ResearchAssistant/App/test_app.py
new file mode 100644
index 00000000..6403b104
--- /dev/null
+++ b/ResearchAssistant/App/test_app.py
@@ -0,0 +1,575 @@
+import json
+import os
+from unittest.mock import MagicMock, Mock, patch
+from flask import Flask
+import pytest
+import urllib
+
+from app import (extract_value, fetchUserGroups,
+                 formatApiResponseNoStreaming, formatApiResponseStreaming,
+                 generateFilterString, is_chat_model, parse_multi_columns,
+                 prepare_body_headers_with_data, should_use_data,
+                 stream_with_data, draft_document_generate)
+
+AZURE_SEARCH_SERVICE = os.environ.get("AZURE_SEARCH_SERVICE", "")
+AZURE_OPENAI_KEY = os.environ.get("AZURE_OPENAI_KEY", "")
+AZURE_SEARCH_PERMITTED_GROUPS_COLUMN = os.environ.get(
+    "AZURE_SEARCH_PERMITTED_GROUPS_COLUMN", ""
+)
+
+
+def test_parse_multi_columns():
+    assert parse_multi_columns("a|b|c") == ["a", "b", "c"]
+    assert parse_multi_columns("a,b,c") == ["a", "b", "c"]
+
+
+@patch("requests.get")
+def test_success_single_page(mock_get):
+    # Mock response for a single page of groups
+    mock_get.return_value.status_code = 200
+    mock_get.return_value.json.return_value = {
+        "value": [{"id": "group1"}, {"id": "group2"}]
+    }
+
+    userToken = "valid_token"
+    result = fetchUserGroups(userToken)
+    expected = [{"id": "group1"}, {"id": "group2"}]
+    assert result == expected
+
+
+def test_is_chat_model_with_gpt4():
+    with patch("app.AZURE_OPENAI_MODEL_NAME", "gpt-4"):
+        assert is_chat_model() is True
+
+
+def test_is_chat_model_with_gpt35_turbo_4k():
+    with patch("app.AZURE_OPENAI_MODEL_NAME", "gpt-35-turbo-4k"):
+        assert is_chat_model() is True
+
+
+def test_is_chat_model_with_gpt35_turbo_16k():
+    with patch("app.AZURE_OPENAI_MODEL_NAME", "gpt-35-turbo-16k"):
+        assert is_chat_model() is True
+
+
+def test_is_chat_model_with_other_model():
+    with patch("app.AZURE_OPENAI_MODEL_NAME", "some-other-model"):
+        assert is_chat_model() is False
+
+
+def test_should_use_data_with_service_and_key():
+    with patch("app.AZURE_SEARCH_SERVICE", "my-service"):
+        with patch("app.AZURE_SEARCH_KEY", "my-key"):
+            with patch("app.DEBUG_LOGGING", False):
+                assert should_use_data() is True
+
+
+def test_should_use_data_with_service_no_key():
+    with patch("app.AZURE_SEARCH_SERVICE", "my-service"):
+        with patch("app.AZURE_SEARCH_KEY", None):
+            assert should_use_data() is False
+
+
+def test_should_use_data_with_key_no_service():
+    with patch("app.AZURE_SEARCH_SERVICE", None):
+        with patch("app.AZURE_SEARCH_KEY", "my-key"):
+            assert should_use_data() is False
+
+
+def test_should_use_data_with_neither():
+    with patch("app.AZURE_SEARCH_SERVICE", None):
+        with patch("app.AZURE_SEARCH_KEY", None):
+            assert should_use_data() is False
+
+
+def test_should_use_data_with_debug_logging():
+    with patch("app.AZURE_SEARCH_SERVICE", "my-service"):
+        with patch("app.AZURE_SEARCH_KEY", "my-key"):
+            with patch("app.DEBUG_LOGGING", True):
+                with patch("logging.debug") as mock_debug:
+                    assert should_use_data() is True
+                    mock_debug.assert_called_once_with("Using Azure Cognitive Search")
+
+
+@patch("requests.get")
+def test_success_multiple_pages(mock_get):
+    # Mock response for multiple pages of groups
+    mock_get.side_effect = [
+        _mock_response(
+            200,
+            {
+                "value": [{"id": "group1"}, {"id": "group2"}],
+                "@odata.nextLink": "https://next.page",
+            },
+        ),
+        _mock_response(200, {"value": [{"id": "group3"}]}),
+    ]
+
+    userToken = "valid_token"
+    result = fetchUserGroups(userToken)
+    expected = [{"id": "group1"}, {"id": "group2"}, {"id": "group3"}]
+    assert result == expected
+
+
+@patch("requests.get")
+def test_non_200_status_code(mock_get):
+    # Mock response with a 403 Forbidden error
+    mock_get.return_value.status_code = 403
+    mock_get.return_value.text = "Forbidden"
+
+    userToken = "valid_token"
+    result = fetchUserGroups(userToken)
+    expected = []
+    assert result == expected
+
+
+@patch("requests.get")
+def test_exception_handling(mock_get):
+    # Mock an exception when making the request
+    mock_get.side_effect = Exception("Network error")
+
+    userToken = "valid_token"
+    result = fetchUserGroups(userToken)
+    expected = []
+    assert result == expected
+
+
+@patch("requests.get")
+def test_no_groups_found(mock_get):
+    # Mock response with no groups found
+    mock_get.return_value.status_code = 200
+    mock_get.return_value.json.return_value = {"value": []}
+
+    userToken = "valid_token"
+    result = fetchUserGroups(userToken)
+    expected = []
+    assert result == expected
+
+
+def _mock_response(status_code, json_data):
+    """Helper method to create a mock response object."""
+    mock_resp = Mock()
+    mock_resp.status_code = status_code
+    mock_resp.json.return_value = json_data
+    return mock_resp
+
+
+@patch("app.fetchUserGroups")
+def test_generateFilterString(mock_fetchUserGroups):
+    mock_fetchUserGroups.return_value = [{"id": "1"}, {"id": "2"}]
+    userToken = "fake_token"
+
+    filter_string = generateFilterString(userToken)
+    assert filter_string == "None/any(g:search.in(g, '1, 2'))"
+
+
+def test_prepare_body_headers_with_data():
+    # Create a mock request
+    mock_request = MagicMock()
+    mock_request.json = {"messages": ["Hello, world!"], "index_name": "grants"}
+    mock_request.headers = {"X-MS-TOKEN-AAD-ACCESS-TOKEN": "mock_token"}
+
+    with patch("app.AZURE_OPENAI_TEMPERATURE", 0.7), patch(
+        "app.AZURE_OPENAI_MAX_TOKENS", 100
+    ), patch("app.AZURE_OPENAI_TOP_P", 0.9), patch(
+        "app.AZURE_SEARCH_SERVICE", "my-service"
+    ), patch(
+        "app.AZURE_SEARCH_KEY", "my-key"
+    ), patch(
+        "app.DEBUG_LOGGING", True
+    ), patch(
+        "app.AZURE_SEARCH_PERMITTED_GROUPS_COLUMN", "group_column"
+    ), patch(
+        "app.AZURE_SEARCH_ENABLE_IN_DOMAIN", "true"
+    ), patch(
+        "app.AZURE_SEARCH_TOP_K", 5
+    ), patch(
+        "app.AZURE_SEARCH_STRICTNESS", 1
+    ):
+
+        body, headers = prepare_body_headers_with_data(mock_request)
+        print("indexName", body["dataSources"][0]["parameters"])
+        assert body["messages"] == ["Hello, world!"]
+        assert body["temperature"] == 0.7
+        assert body["max_tokens"] == 100
+        assert body["top_p"] == 0.9
+        assert body["dataSources"]
+        assert body["dataSources"][0]["type"] == "AzureCognitiveSearch"
+        assert (
+            body["dataSources"][0]["parameters"]["endpoint"]
+            == "https://my-service.search.windows.net"
+        )
+        assert body["dataSources"][0]["parameters"]["key"] == "my-key"
+        assert body["dataSources"][0]["parameters"]["inScope"] is True
+        assert body["dataSources"][0]["parameters"]["topNDocuments"] == 5
+        assert body["dataSources"][0]["parameters"]["strictness"] == 1
+
+        assert headers["Content-Type"] == "application/json"
+        assert headers["x-ms-useragent"] == "GitHubSampleWebApp/PublicAPI/3.0.0"
+
+
+def test_invalid_datasource_type():
+    mock_request = MagicMock()
+    mock_request.json = {"messages": ["Hello, world!"], "index_name": "grants"}
+
+    with patch("app.DATASOURCE_TYPE", "InvalidType"):
+        with pytest.raises(Exception) as exc_info:
+            prepare_body_headers_with_data(mock_request)
+        assert "DATASOURCE_TYPE is not configured or unknown: InvalidType" in str(
+            exc_info.value
+        )
+
+
+# stream_with_data function
+def mock_format_as_ndjson(data):
+    # Ensure data is in a JSON serializable format (like a list or dict)
+    if isinstance(data, set):
+        data = list(data)  # Convert set to list
+    return json.dumps(data)
+
+
+def test_stream_with_data_azure_success():
+    body = {
+        "messages": [
+            {
+                "id": "0e29210d-5584-38df-df76-2dfb40147ee7",
+                "role": "user",
+                "content": "influenza and its effets ",
+                "date": "2025-01-09T04:42:25.896Z",
+            },
+            {
+                "id": "ab42add2-0fba-d6bb-47c0-5d11b7cdb83a",
+                "role": "user",
+                "content": "influenza and its effectd",
+                "date": "2025-01-09T10:14:11.638Z",
+            },
+            {
+                "id": "1f6dc8e2-c5fe-ce77-b28c-5ec9ba80e94d",
+                "role": "user",
+                "content": "influenza and its effects",
+                "date": "2025-01-09T10:34:15.187Z",
+            },
+        ],
+        "temperature": 0.0,
+        "max_tokens": 1000,
+        "top_p": 1.0,
+        "stop": "None",
+        "stream": True,
+        "dataSources": [
+            {
+                "type": "AzureCognitiveSearch",
+                "parameters": {
+                    "endpoint": "https://ututut-cs.search.windows.net",
+                    "key": "",
+                    "indexName": "articlesindex",
+                    "fieldsMapping": {
+                        "contentFields": ["content"],
+                        "titleField": "title",
+                        "urlField": "publicurl",
+                        "filepathField": "chunk_id",
+                        "vectorFields": ["titleVector", "contentVector"],
+                    },
+                    "inScope": False,
+                    "topNDocuments": "5",
+                    "queryType": "vectorSemanticHybrid",
+                    "semanticConfiguration": "my-semantic-config",
+                    "roleInformation": "You are an AI assistant that helps people find information.",
+                    "filter": "None",
+                    "strictness": 3,
+                    "embeddingDeploymentName": "text-embedding-ada-002",
+                },
+            }
+        ],
+    }
+    headers = {
+        "Content-Type": "application/json",
+        "api-key": "",
+        "x-ms-useragent": "GitHubSampleWebApp/PublicAPI/3.0.0",
+    }
+    history_metadata = {}
+
+    with patch("requests.Session.post") as mock_post:
+        mock_response = MagicMock()
+        mock_response.iter_lines.return_value = [
+            b'data: {"id":"1","model":"gpt-35-turbo-16k","created":1736397875,"object":"extensions.chat.completion.chunk","choices":[{"index":0,"delta":{"context":{"messages":[{"role":"tool","content":"hello","end_turn":false}]}},"end_turn":false,"finish_reason":"None"}]}'
+        ]
+        mock_response.headers = {"apim-request-id": "test-request-id"}
+        mock_post.return_value.__enter__.return_value = mock_response
+
+        with patch("app.format_as_ndjson", side_effect=mock_format_as_ndjson):
+            results = list(
+                stream_with_data(
+                    body, headers, "https://mock-endpoint.com", history_metadata
+                )
+            )  # Convert generator to a list
+            print(results, "result test case")
+            assert len(results) == 1
+
+
+# Mock constants
+USE_AZURE_AI_STUDIO = "true"
+AZURE_OPENAI_PREVIEW_API_VERSION = "2023-06-01-preview"
+DEBUG_LOGGING = False
+AZURE_SEARCH_SERVICE = os.environ.get("AZURE_SEARCH_SERVICE", "mysearchservice")
+
+
+def test_stream_with_data_azure_error():
+    body = {
+        "messages": [
+            {
+                "id": "0e29210d-5584-38df-df76-2dfb40147ee7",
+                "role": "user",
+                "content": "influenza and its effets ",
+                "date": "2025-01-09T04:42:25.896Z",
+            },
+            {
+                "id": "ab42add2-0fba-d6bb-47c0-5d11b7cdb83a",
+                "role": "user",
+                "content": "influenza and its effectd",
+                "date": "2025-01-09T10:14:11.638Z",
+            },
+            {
+                "id": "1f6dc8e2-c5fe-ce77-b28c-5ec9ba80e94d",
+                "role": "user",
+                "content": "influenza and its effects",
+                "date": "2025-01-09T10:34:15.187Z",
+            },
+        ],
+        "temperature": 0.0,
+        "max_tokens": 1000,
+        "top_p": 1.0,
+        "stop": "None",
+        "stream": True,
+        "dataSources": [
+            {
+                "type": "AzureCognitiveSearch",
+                "parameters": {
+                    "endpoint": "https://ututut-cs.search.windows.net",
+                    "key": "",
+                    "indexName": "articlesindex",
+                    "fieldsMapping": {
+                        "contentFields": ["content"],
+                        "titleField": "title",
+                        "urlField": "publicurl",
+                        "filepathField": "chunk_id",
+                        "vectorFields": ["titleVector", "contentVector"],
+                    },
+                    "inScope": False,
+                    "topNDocuments": "5",
+                    "queryType": "vectorSemanticHybrid",
+                    "semanticConfiguration": "my-semantic-config",
+                    "roleInformation": "You are an AI assistant that helps people find information.",
+                    "filter": "None",
+                    "strictness": 3,
+                    "embeddingDeploymentName": "text-embedding-ada-002",
+                },
+            }
+        ],
+    }
+
+    if USE_AZURE_AI_STUDIO.lower() == "true":
+        body = body
+    headers = {
+        "Content-Type": "application/json",
+        "api-key": "",
+        "x-ms-useragent": "GitHubSampleWebApp/PublicAPI/3.0.0",
+    }
+    history_metadata = {}
+
+    with patch("requests.Session.post") as mock_post:
+        # if USE_AZURE_AI_STUDIO.lower() == "true":
+        #     body = mock_body
+        mock_response = MagicMock()
+        mock_response.iter_lines.return_value = [
+            b'data: {"id":"1","model":"gpt-35-turbo-16k","created":1736397875,"object":"extensions.chat.completion.chunk","choices":[{"index":0,"delta":{"context":{"messages":[{"role":"tool","content":"hello","end_turn":false}]}},"end_turn":false,"finish_reason":"None"}]}'
+        ]
+        mock_response.headers = {"apim-request-id": "test-request-id"}
+        mock_post.return_value.__enter__.return_value = mock_response
+
+        with patch("app.format_as_ndjson", side_effect=mock_format_as_ndjson):
+            results = list(
+                stream_with_data(
+                    body, headers, "https://mock-endpoint.com", history_metadata
+                )
+            )  # Convert generator to a list
+            print(results, "result test case")
+            assert len(results) == 1
+
+
+def test_formatApiResponseNoStreaming():
+    rawResponse = {
+        "id": "1",
+        "model": "gpt-3",
+        "created": 123456789,
+        "object": "response",
+        "choices": [
+            {
+                "message": {
+                    "context": {"messages": [{"content": "Hello from tool"}]},
+                    "content": "Hello from assistant",
+                }
+            }
+        ],
+    }
+    response = formatApiResponseNoStreaming(rawResponse)
+    assert "choices" in response
+    assert response["choices"][0]["messages"][0]["content"] == "Hello from tool"
+
+
+def test_formatApiResponseStreaming():
+    rawResponse = {
+        "id": "1",
+        "model": "gpt-3",
+        "created": 123456789,
+        "object": "response",
+        "choices": [{"delta": {"role": "assistant", "content": "Hello"}}],
+    }
+
+    response = formatApiResponseStreaming(rawResponse)
+
+    # Print response to debug
+    print(response)  # Optional for debugging, remove in production
+
+    assert "choices" in response
+    assert "messages" in response["choices"][0]
+    assert len(response["choices"][0]["messages"]) == 1
+
+    # Check if the content is included under the correct structure
+    delta_content = response["choices"][0]["messages"][0]["delta"]
+    assert "role" in delta_content  # Check for role
+    assert (
+        "content" not in delta_content
+    )  # content should not be present as per current logic
+
+
+def test_extract_value():
+    text = "'code': 'content_filter', 'status': '400'"
+    assert extract_value("code", text) == "content_filter"
+    assert extract_value("status", text) == "400"
+    assert extract_value("unknown", text) == "N/A"
+
+
+app = Flask(__name__)
+
+
+app.add_url_rule("/draft_document/generate_section", "draft_document_generate", draft_document_generate, methods=["POST"])
+
+
+# Helper to create a mock response
+class MockResponse:
+    def __init__(self, json_data, status_code):
+        self.json_data = json_data
+        self.status_code = status_code
+
+    def read(self):
+        return json.dumps(self.json_data).encode('utf-8')
+
+    def getcode(self):
+        return self.status_code
+
+
+@pytest.fixture
+def client():
+    with app.test_client() as client:
+        yield client
+
+
+# Test the successful response case
+@patch("urllib.request.urlopen")
+@patch("os.environ.get")
+def test_draft_document_generate_success(mock_os_environ, mock_urlopen, client):
+    mock_os_environ.side_effect = lambda key: {
+        "AI_STUDIO_DRAFT_FLOW_ENDPOINT": "https://fakeurl.com",
+        "AI_STUDIO_DRAFT_FLOW_API_KEY": "fakeapikey",
+        "AI_STUDIO_DRAFT_FLOW_DEPLOYMENT_NAME": "fake_deployment_name"
+    }.get(key)
+
+    # Mock the successful API response
+    mock_urlopen.return_value = MockResponse({"reply": "Generated content for section."}, 200)
+
+    # Sample input payload
+    payload = {
+        "grantTopic": "Artificial Intelligence",
+        "sectionTitle": "Introduction",
+        "sectionContext": ""
+    }
+
+    response = client.post("/draft_document/generate_section", json=payload)
+
+    # Assertions
+    assert response.status_code == 200
+    response_json = response.get_json()
+    assert "content" in response_json
+    assert response_json["content"] == "Generated content for section."
+
+
+# Test the scenario where "sectionContext" is provided
+@patch("urllib.request.urlopen")
+@patch("os.environ.get")
+def test_draft_document_generate_with_context(mock_os_environ, mock_urlopen, client):
+    mock_os_environ.side_effect = lambda key: {
+        "AI_STUDIO_DRAFT_FLOW_ENDPOINT": "https://fakeurl.com",
+        "AI_STUDIO_DRAFT_FLOW_API_KEY": "fakeapikey",
+        "AI_STUDIO_DRAFT_FLOW_DEPLOYMENT_NAME": "fake_deployment_name"
+    }.get(key)
+
+    # Mock the successful API response
+    mock_urlopen.return_value = MockResponse({"reply": "Generated content with context."}, 200)
+
+    payload = {
+        "grantTopic": "Quantum Computing",
+        "sectionTitle": "Background",
+        "sectionContext": "The section should explain the significance of quantum computing."
+    }
+
+    response = client.post("/draft_document/generate_section", json=payload)
+
+    # Assertions
+    assert response.status_code == 200
+    response_json = response.get_json()
+    assert "content" in response_json
+    assert response_json["content"] == "Generated content with context."
+
+
+@pytest.fixture
+def clients():
+    app = Flask(__name__)
+    app.route('/draft_document/generate_section', methods=['POST'])(draft_document_generate)
+    client = app.test_client()
+    yield client
+
+
+@patch("urllib.request.urlopen")
+@patch("os.environ.get")
+def test_draft_document_generate_http_error(mock_env_get, mock_urlopen, client):
+    # Mock environment variables
+    mock_env_get.side_effect = lambda key: {
+        "AI_STUDIO_DRAFT_FLOW_ENDPOINT": "http://mock_endpoint",
+        "AI_STUDIO_DRAFT_FLOW_API_KEY": "mock_api_key",
+        "AI_STUDIO_DRAFT_FLOW_DEPLOYMENT_NAME": "mock_deployment"
+    }.get(key)
+
+    # Mock urllib.request.urlopen to raise an HTTPError
+    error_response = json.dumps({"error": {"message": "content_filter", "code": "400"}}).encode('utf-8')
+    mock_urlopen.side_effect = urllib.error.HTTPError(
+        url="http://mock_endpoint",
+        code=400,
+        msg="Bad Request",
+        hdrs=None,
+        fp=MagicMock(read=MagicMock(return_value=error_response))
+    )
+
+    # Mock request data
+    request_data = {
+        "grantTopic": "Climate Change Research",
+        "sectionTitle": "Introduction",
+        "sectionContext": "This research focuses on reducing carbon emissions."
+    }
+
+    response = client.post(
+        "/draft_document/generate_section",
+        data=json.dumps(request_data),
+        content_type="application/json",
+    )
+
+    assert response.status_code == 200