diff --git a/.github/workflows/ci.yaml b/.github/workflows/ci.yaml
index fd46fcd..c129ea9 100644
--- a/.github/workflows/ci.yaml
+++ b/.github/workflows/ci.yaml
@@ -121,7 +121,27 @@ jobs:
           ./tests/wait_hserver.sh
           sleep 5
 
-      - name: Run examples
+      - name: Run hstream examples
         run: |
           python3 -m pip install hstreamdb
           python3 examples/py/snippets/guides.py
+
+  test-python-kafka-examples:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+
+      - name: Start cluster
+        run: |
+          docker compose -f ./assets/quick-start-kafka-lite.yaml up -d
+          sleep 5
+          ./tests/wait_hstream_kafka.sh
+          sleep 5
+
+      - name: Run hstream kafka examples
+        run: |
+          python3 -m pip install kafka-python
+          python3 kafka-examples/python/snippets/kafka_python.py
+
+          python3 -m pip install confluent-kafka
+          python3 kafka-examples/python/snippets/confluent_kafka_python.py
diff --git a/assets/quick-start-kafka-lite.yaml b/assets/quick-start-kafka-lite.yaml
new file mode 100644
index 0000000..9b085c3
--- /dev/null
+++ b/assets/quick-start-kafka-lite.yaml
@@ -0,0 +1,74 @@
+version: "3.5"
+
+services:
+  hserver:
+    image: hstreamdb/hstream:latest
+    depends_on:
+      - zookeeper
+      - hstore
+    ports:
+      - "127.0.0.1:9092:9092"
+    expose:
+      - 9092
+    networks:
+      - hstream-quickstart
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock
+      - /tmp:/tmp
+      - data_store:/data/store
+    command:
+      - bash
+      - "-c"
+      - |
+        set -e
+        /usr/local/script/wait-for-storage.sh hstore 6440 zookeeper 2181 600 \
+        /usr/local/bin/hstream-server kafka \
+          --bind-address 0.0.0.0 --port 9092 \
+          --gossip-port 6571 \
+          --server-id 100 \
+          --seed-nodes "$$(hostname -I | awk '{print $$1}'):6571" \
+          --advertised-address $$(hostname -I | awk '{print $$1}') \
+          --metastore-uri zk://zookeeper:2181 \
+          --store-config /data/store/logdevice.conf \
+          --store-log-level warning
+
+  hstore:
+    image: hstreamdb/hstream:latest
+    networks:
+      - hstream-quickstart
+    volumes:
+      - data_store:/data/store
+    command:
+      - bash
+      - "-c"
+      - |
+        set -ex
+        # N.B. "enable-dscp-reflection=false" is required for linux kernel which
+        # doesn't support dscp reflection, e.g. centos7.
+        /usr/local/bin/ld-dev-cluster --root /data/store \
+        --use-tcp --tcp-host $$(hostname -I | awk '{print $$1}') \
+        --user-admin-port 6440 \
+        --param enable-dscp-reflection=false \
+        --no-interactive
+
+  zookeeper:
+    image: zookeeper:3.7
+    expose:
+      - 2181
+    networks:
+      - hstream-quickstart
+    volumes:
+      - data_zk_data:/data
+      - data_zk_datalog:/datalog
+
+networks:
+  hstream-quickstart:
+    name: hstream-quickstart
+
+volumes:
+  data_store:
+    name: quickstart_data_store
+  data_zk_data:
+    name: quickstart_data_zk_data
+  data_zk_datalog:
+    name: quickstart_data_zk_datalog
diff --git a/docs/develop-with-kafka-api/_index.md b/docs/develop-with-kafka-api/_index.md
index 69b5f99..396af55 100644
--- a/docs/develop-with-kafka-api/_index.md
+++ b/docs/develop-with-kafka-api/_index.md
@@ -1,5 +1,5 @@
 ---
-order: ['compatibility.md', 'java.md']
+order: ['compatibility.md', 'java.md', 'python.md']
 collapsed: false
 ---
 
diff --git a/docs/develop-with-kafka-api/compatibility.md b/docs/develop-with-kafka-api/compatibility.md
index 735b583..5ecc334 100644
--- a/docs/develop-with-kafka-api/compatibility.md
+++ b/docs/develop-with-kafka-api/compatibility.md
@@ -2,24 +2,31 @@
 
 ## Overview
 
-HStream also supports Kafka API since 0.19.0, so users can connect to HStream using Kafka clients. HStream implements [Kafka protocol](https://kafka.apache.org/protocol.html) underly, so you do not need to change any code in your current Kafka applications, just updating the Kafka URLs in your configurations to point to a HStream cluster, and that is it, then you can start streaming from your Kafka applications to a HStream cluster.
+HStream also supports Kafka API since 0.19.0, so users can connect to HStream
+using Kafka clients. HStream implements
+[Kafka protocol](https://kafka.apache.org/protocol.html) underly, so you do not
+need to change any code in your current Kafka applications, just updating the
+Kafka URLs in your configurations to point to a HStream cluster, and that is it,
+then you can start streaming from your Kafka applications to a HStream cluster.
 
 ::: tip
 
-Refer to [get started with Kafka API](../start/get-started-with-kafka-api.md) to learn how to enable HStream'support of Kafka API.
+Refer to [get started with Kafka API](../start/get-started-with-kafka-api.md) to
+learn how to enable HStream'support of Kafka API.
 
 :::
 
 ## Compatibility with Apache Kafka
 
-HStream supports Apache Kafka version 0.11 and later, and most Kafka clients should be able to auto-negotiate protocol versions.
+HStream supports Apache Kafka version 0.11 and later, and most Kafka clients
+should be able to auto-negotiate protocol versions.
 
 Currenty, the clients below are tested by HStream.
 
 | Language | Kafka Client                                                |
 | -------- | ----------------------------------------------------------- |
 | Java     | [Apache Kafka Java Client](https://github.com/apache/kafka) |
-| Python   | [kafka-python](https://github.com/dpkp/kafka-python)        |
+| Python   | [kafka-python], [confluent-kafka-python]                    |
 | Go       | [franz-go](https://github.com/twmb/franz-go)                |
 | C/C++    | [librdkafka](https://github.com/confluentinc/librdkafka)    |
 
@@ -31,13 +38,18 @@ Recommand using the latest version of each Kafka client
 
 ## Features not supported in Apache Kafka
 
-HStream do not support below Kafka features now(we plan to support them in the later version):
+HStream do not support below Kafka features now(we plan to support them in the
+later version):
 
 - Kafka transactions
 - Quotas in Kafka
 
 ::: tip
 
-The configuration of Kafka brokers is not applicable to HStream, as HStream is a completely different implementation.
+The configuration of Kafka brokers is not applicable to HStream, as HStream is a
+completely different implementation.
 
 :::
+
+[kafka-python]: https://github.com/dpkp/kafka-python
+[confluent-kafka-python]: https://github.com/confluentinc/confluent-kafka-python
diff --git a/docs/develop-with-kafka-api/python.md b/docs/develop-with-kafka-api/python.md
new file mode 100644
index 0000000..439f035
--- /dev/null
+++ b/docs/develop-with-kafka-api/python.md
@@ -0,0 +1,60 @@
+# Develop with Python Kafka client
+
+This guide will show you how to use Python Kafka client to interact with
+HStream. Currenty, we support [kafka-python] and [confluent-kafka].
+
+## Installation
+
+```sh
+# If you want to use kafka-python
+pip install kafka-python
+
+# Or if you want to use confluent-kafka
+pip install confluent-kafka
+```
+
+::: tip
+
+Prefer to use a virtual environment? Check out Python's built-in
+[venv](https://docs.python.org/3/library/venv.html).
+
+:::
+
+## Create a Topic
+
+::: code-group
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/kafka_python.py [kafka-python] common create-topic
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/confluent_kafka_python.py [confluent-kafka] common create-topic
+
+:::
+
+## Produce Records
+
+::: code-group
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/kafka_python.py [kafka-python] common produce
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/confluent_kafka_python.py [confluent-kafka] common produce
+
+:::
+
+## Consume Records
+
+::: code-group
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/kafka_python.py [kafka-python] common consume
+
+<!-- prettier-ignore -->
+@snippet_group kafka-examples/python/snippets/confluent_kafka_python.py [confluent-kafka] common consume
+
+:::
+
+[kafka-python]: https://github.com/dpkp/kafka-python
+[confluent-kafka]: https://github.com/confluentinc/confluent-kafka-python
diff --git a/docs/develop/_index.md b/docs/develop/_index.md
index 28531c5..3a1bccf 100644
--- a/docs/develop/_index.md
+++ b/docs/develop/_index.md
@@ -3,4 +3,4 @@ order: ['write', 'receive', 'process', 'ingest-and-distribute']
 collapsed: false
 ---
 
-Develope
+Develop
diff --git a/embed.sh b/embed.sh
index 6c41452..1c49ad0 100755
--- a/embed.sh
+++ b/embed.sh
@@ -4,4 +4,5 @@
 if [[ "$EMBED_ME_ARGS" == "" ]]; then
   ./include_snippets.py --snippets-root . --file-pattern "docs/**/write/*.md"
   ./include_snippets.py --snippets-root . --file-pattern "docs/**/receive/*.md"
+  ./include_snippets.py --snippets-root . --file-pattern "docs/**/python.md"
 fi
diff --git a/include_snippets.py b/include_snippets.py
index 81a0377..77c9253 100755
--- a/include_snippets.py
+++ b/include_snippets.py
@@ -2,10 +2,10 @@
 
 import os
 
-FILE_TYPES = {".py": "python", "cpp": "cpp"}
+FILE_TYPES = {".py": ("python", "[Python]"), ".cpp": ("cpp", "[C++]")}
 
 
-def run_snippet_cmd(snippets_root, snippet_file, *labels):
+def run_snippet_cmd(snippets_root, snippet_file, *labels, code_group=None):
     lines = [[] for _ in labels]
     with open(os.path.join(snippets_root, snippet_file), "r") as f:
         idx = None
@@ -24,12 +24,22 @@ def run_snippet_cmd(snippets_root, snippet_file, *labels):
                     lines[idx].append(line)
     # TODO: strip indent
     extension = os.path.splitext(snippet_file)[1]
-    ft = FILE_TYPES[extension]
+    (ft, _code_group) = FILE_TYPES[extension]
+    code_group = _code_group if code_group is None else code_group
     blocks = "\n\n\n".join("".join(xs).strip() for xs in lines)
-    return f"```{ft} [Python]\n{blocks}\n```\n"
+    return f"```{ft} {code_group}\n{blocks}\n```\n"
 
 
-support_commands = [("@snippet", run_snippet_cmd)]
+def run_snippet_cmd_group(snippets_root, snippet_file, group, *labels):
+    return run_snippet_cmd(
+        snippets_root, snippet_file, *labels, code_group=group
+    )
+
+
+support_commands = [
+    ("@snippet", run_snippet_cmd),
+    ("@snippet_group", run_snippet_cmd_group),
+]
 
 
 def parse_commands(lines):
diff --git a/kafka-examples/python/snippets/confluent_kafka_python.py b/kafka-examples/python/snippets/confluent_kafka_python.py
new file mode 100644
index 0000000..377d038
--- /dev/null
+++ b/kafka-examples/python/snippets/confluent_kafka_python.py
@@ -0,0 +1,106 @@
+# [common]
+
+import os
+from confluent_kafka import Producer, Consumer
+from confluent_kafka.admin import AdminClient, NewTopic
+
+# NOTE: Replace with your own host and port
+host = os.getenv("GUIDE_HOST", "127.0.0.1")
+port = os.getenv("GUIDE_PORT", 9092)
+
+topic_name = "my_topic"
+group_id = "confluent_kafka_group"
+
+conf = {
+    "bootstrap.servers": host + ":" + str(port),
+    "client.id": "confluent_kafka_client",
+}
+# [common]
+
+
+# [create-topic]
+def create_topic():
+    admin = AdminClient(conf)
+    new_topic = NewTopic(topic_name, num_partitions=1, replication_factor=1)
+    admin.create_topics([new_topic])
+
+
+# [create-topic]
+
+
+# [produce]
+def produce():
+    def acked(err, msg):
+        if err is not None:
+            print(f"Failed to deliver message: {msg}: {err}")
+        else:
+            print(
+                f"Message produced: offset={msg.offset()}, "
+                f'key="{msg.key().decode()}", '
+                f'value="{msg.value().decode()}"'
+            )
+
+    producer = Producer(conf)
+    for i in range(5):
+        producer.produce(
+            topic_name,
+            key=b"key " + str(i).encode(),
+            value=b"hello, hstream " + str(i).encode(),
+            on_delivery=acked,
+        )
+    producer.flush()
+
+
+# [produce]
+
+
+# [consume]
+def consume():
+    consumer = Consumer(
+        {
+            **conf,
+            "group.id": group_id,
+            "auto.offset.reset": "smallest",
+            "enable.auto.commit": "false",
+            "isolation.level": "read_uncommitted",
+        }
+    )
+    consumer.subscribe([topic_name])
+    i = 0
+    try:
+        while True:
+            msg = consumer.poll(1.0)
+            if msg is None:
+                # Initial message consumption may take up to
+                # `session.timeout.ms` for the consumer group to
+                # rebalance and start consuming
+                print("Waiting...")
+            elif msg.error():
+                print(f"ERROR: {msg.error()}")
+            else:
+                # Extract the (optional) key and value, and print.
+                print(
+                    f"Consumed topic {msg.topic()}: "
+                    f'key="{msg.key().decode()}", '
+                    f'value="{msg.value().decode()}"'
+                )
+                i += 1
+            if i >= 5:
+                break
+    except KeyboardInterrupt:
+        pass
+    finally:
+        consumer.close()
+
+
+# [consume]
+
+
+if __name__ == "__main__":
+    try:
+        create_topic()
+        produce()
+        consume()
+    finally:
+        admin = AdminClient(conf)
+        admin.delete_topics([topic_name])
diff --git a/kafka-examples/python/snippets/kafka_python.py b/kafka-examples/python/snippets/kafka_python.py
new file mode 100644
index 0000000..26ced10
--- /dev/null
+++ b/kafka-examples/python/snippets/kafka_python.py
@@ -0,0 +1,72 @@
+# [common]
+
+import os
+from kafka.admin import NewTopic
+from kafka import KafkaAdminClient, KafkaConsumer, KafkaProducer
+
+# NOTE: Replace with your own host and port
+host = os.getenv("GUIDE_HOST", "127.0.0.1")
+port = os.getenv("GUIDE_PORT", 9092)
+addr = host + ":" + str(port)
+topic_name = "my_topic"
+
+# [common]
+
+
+# [create-topic]
+def create_topic():
+    admin = KafkaAdminClient(bootstrap_servers=addr)
+    topic = NewTopic(name=topic_name, num_partitions=1, replication_factor=1)
+    admin.create_topics([topic])
+
+
+# [create-topic]
+
+
+# [produce]
+def produce():
+    producer = KafkaProducer(
+        bootstrap_servers=addr,
+        acks="all",
+        linger_ms=100,
+    )
+    futures = [
+        producer.send(topic_name, b"hello, hstream " + str(i).encode())
+        for i in range(5)
+    ]
+    for future in futures:
+        response = future.get(timeout=10)
+        print("Producer response:", response)
+
+
+# [produce]
+
+
+# [consume]
+def consume():
+    consumer = KafkaConsumer(
+        topic_name,
+        bootstrap_servers=addr,
+        auto_offset_reset="earliest",
+        enable_auto_commit=False,
+        fetch_max_wait_ms=1000,
+    )
+    i = 0
+    for msg in consumer:
+        print("Consumer response", msg)
+        i += 1
+        if i >= 5:
+            consumer.close()
+
+
+# [consume]
+
+
+if __name__ == "__main__":
+    try:
+        create_topic()
+        produce()
+        consume()
+    finally:
+        admin = KafkaAdminClient(bootstrap_servers=addr)
+        admin.delete_topics([topic_name])
diff --git a/tests/wait_hstream_kafka.sh b/tests/wait_hstream_kafka.sh
new file mode 100755
index 0000000..0ab5475
--- /dev/null
+++ b/tests/wait_hstream_kafka.sh
@@ -0,0 +1,14 @@
+#!/bin/bash
+set -e
+
+timeout=120
+container_name=wait_hserver
+
+docker run -td --rm --name $container_name --network host hstreamdb/hstream bash
+until (docker exec -t $container_name hstream-kafka-cli --port 9092 node status); do
+    >&2 echo "Waiting for 127.0.0.1:9092 ..."
+    sleep 1
+    timeout=$((timeout - 1))
+    [ $timeout -le 0 ] && echo "Timeout!" && docker kill $container_name && exit 1;
+done
+docker kill $container_name