diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index eb0cf5e..8673d3c 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -36,3 +36,13 @@ jobs:
         export PYTHONPATH=${SPARK_HOME}/python:${SPARK_HOME}/python/lib/py4j-0.10.7-src.zip:${PYTHONPATH}
         export PATH=${PATH}:${SPARK_HOME}/bin:${SPARK_HOME}/sbin
         make test
+    - name: Convert notebook to python script
+      run: |
+        make py
+        mkdir -p ima/scripts/
+        mv algorithm.py ima/scripts
+    - uses: actions/upload-artifact@v2
+      with:
+        name: algorithm
+        path: ima/scripts/algorithm.py
+
diff --git a/Makefile b/Makefile
index 8ec4459..62c6338 100644
--- a/Makefile
+++ b/Makefile
@@ -2,6 +2,7 @@ spark_version := 2.4.8
 hadoop_version := 2.7
 spark_home := spark-${spark_version}-bin-hadoop${hadoop_version}
 spark_tgz_url := https://downloads.apache.org/spark/spark-${spark_version}/${spark_home}.tgz
+ima_notebook := algorithm.ipynb
 
 venv: requirements.txt
 	test -d venv || python3 -m venv venv
@@ -21,3 +22,7 @@ flake8:	venv
 
 test:	venv
 	. venv/bin/activate; PYTHONPATH=${PYTHONPATH}:etl/ pytest --cov etl tests/
+
+py:	venv
+	# nbconvert output is saved as <basename>.py
+	. venv/bin/activate; jupyter nbconvert ${ima_notebook} --to script
diff --git a/requirements.txt b/requirements.txt
index 8de2ac0..b59b67e 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -2,3 +2,4 @@ pytest==6.2.2
 pytest-spark==0.6.0
 pytest-cov==2.10.1
 flake8==3.8.4
+jupyter==1.0.0