diff --git a/Cargo.lock b/Cargo.lock
index 3222798..e3b2840 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -579,6 +579,20 @@ dependencies = [
  "parking_lot_core",
 ]
 
+[[package]]
+name = "demo"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "env_logger",
+ "hdf5-metno",
+ "log",
+ "ndarray",
+ "proto",
+ "tokio",
+ "tonic",
+]
+
 [[package]]
 name = "dirs"
 version = "3.0.2"
diff --git a/Cargo.toml b/Cargo.toml
index ca5478e..e0ac3b8 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,5 +1,6 @@
 [workspace]
-members = [
+members = [ 
+    "rs/demo",
     "rs/aggregator", 
     "rs/cli", 
     "rs/compression", 
diff --git a/py/create_1m_hdf5.py b/py/create_1m_hdf5.py
new file mode 100644
index 0000000..fcae8e9
--- /dev/null
+++ b/py/create_1m_hdf5.py
@@ -0,0 +1,32 @@
+import h5py
+import numpy as np
+
+def create_hdf5(input_path, output_path, batch_size=100000):
+    # Create HDF5 file
+    with h5py.File(output_path, 'w') as hf:
+        # Initialize dataset with unknown size
+        dataset = hf.create_dataset('embeddings', 
+                                  shape=(0, 768), 
+                                  maxshape=(None, 768),
+                                  dtype='float32')
+        
+        # Process file in batches
+        with open(input_path, 'r') as f:
+            while True:
+                # Read batch of lines
+                lines = [f.readline() for _ in range(batch_size)]
+                if not lines[0]:  # End of file
+                    break
+                    
+                # Parse lines to numpy array
+                embeddings = np.array([np.fromstring(line.strip('[]\n'), sep=',') 
+                                     for line in lines if line.strip()])
+                
+                # Resize dataset and append new data
+                dataset.resize(dataset.shape[0] + embeddings.shape[0], axis=0)
+                dataset[-embeddings.shape[0]:] = embeddings
+
+if __name__ == '__main__':
+    input_file = '/mnt/muopdb/raw/1m_embeddings.txt'
+    output_file = '/mnt/muopdb/1m_embeddings.hdf5'
+    create_hdf5(input_file, output_file)
diff --git a/py/demo_insert_search.py b/py/demo_insert_search.py
index 2cdb12a..ad3f534 100644
--- a/py/demo_insert_search.py
+++ b/py/demo_insert_search.py
@@ -1,166 +1,56 @@
 import muopdb_client as mp
 import google.generativeai as genai
 import os
-
-SENTENCES = [
-    "The sun rises in the east and sets in the west.",
-    "Rainforests are home to millions of species of plants and animals.",
-    "Climate change is causing glaciers to melt at an alarming rate.",
-    "Bees play a crucial role in pollinating crops.",
-    "The ocean covers more than 70% of the Earth's surface.",
-    "Planting trees can help reduce carbon dioxide in the atmosphere.",
-    "Hurricanes are becoming more intense due to global warming.",
-    "The Grand Canyon is one of the most breathtaking natural wonders.",
-    "Recycling helps reduce waste in landfills.",
-    "Solar energy is a clean and renewable source of power.",
-    "Artificial intelligence is transforming industries worldwide.",
-    "Smartphones have become an essential part of modern life.",
-    "Electric vehicles are gaining popularity as an eco-friendly alternative.",
-    "Social media platforms connect people across the globe.",
-    "Cybersecurity is critical to protecting personal data online.",
-    "Virtual reality allows users to experience immersive environments.",
-    "3D printing is revolutionizing manufacturing processes.",
-    "Blockchain technology is the backbone of cryptocurrencies.",
-    "Robots are increasingly being used in healthcare and surgery.",
-    "The internet has made information accessible to billions of people.",
-    "Regular exercise improves both physical and mental health.",
-    "Drinking enough water is essential for staying hydrated.",
-    "A balanced diet includes fruits, vegetables, and whole grains.",
-    "Meditation can help reduce stress and anxiety.",
-    "Getting enough sleep is crucial for overall well-being.",
-    "Smoking is a leading cause of lung cancer.",
-    "Yoga combines physical postures with mindfulness practices.",
-    "Vaccines have eradicated many deadly diseases.",
-    "Walking for 30 minutes a day can boost cardiovascular health.",
-    "Mental health is just as important as physical health.",
-    "Paris is known as the 'City of Love' and is famous for the Eiffel Tower.",
-    "Japan is a country that blends tradition with modernity.",
-    "The Great Wall of China is one of the Seven Wonders of the World.",
-    "Italy is renowned for its delicious cuisine and rich history.",
-    "Traveling broadens your perspective and introduces you to new cultures.",
-    "The Northern Lights are a spectacular natural phenomenon.",
-    "India is a diverse country with many languages and traditions.",
-    "Australia is home to unique wildlife like kangaroos and koalas.",
-    "The pyramids of Egypt are a testament to ancient engineering.",
-    "New York City is often called 'The City That Never Sleeps.'",
-    "Reading books enhances vocabulary and critical thinking skills.",
-    "Lifelong learning is key to personal and professional growth.",
-    "Online courses have made education more accessible.",
-    "Mathematics is the foundation of many scientific disciplines.",
-    "Learning a new language can open up career opportunities.",
-    "History teaches us valuable lessons about the past.",
-    "Creativity is just as important as technical skills in many fields.",
-    "Teachers play a vital role in shaping young minds.",
-    "STEM education focuses on science, technology, engineering, and math.",
-    "Libraries are a valuable resource for knowledge and research.",
-    "Pizza originated in Italy and is now popular worldwide.",
-    "Sushi is a traditional Japanese dish made with rice and seafood.",
-    "Cooking at home is often healthier than eating out.",
-    "Chocolate is made from cocoa beans and is loved by many.",
-    "Spices like turmeric and cumin add flavor to dishes.",
-    "Breakfast is often called the most important meal of the day.",
-    "Vegetarian diets exclude meat but include plant-based foods.",
-    "Baking requires precise measurements and techniques.",
-    "Coffee is one of the most consumed beverages in the world.",
-    "Food waste is a major issue that needs to be addressed.",
-    "Leonardo da Vinci painted the famous Mona Lisa.",
-    "Music has the power to evoke emotions and bring people together.",
-    "Shakespeare is considered one of the greatest playwrights in history.",
-    "Movies are a popular form of entertainment worldwide.",
-    "The Beatles revolutionized the music industry in the 1960s.",
-    "Photography captures moments and tells stories visually.",
-    "Broadway is known for its spectacular theatrical performances.",
-    "Graffiti is a form of street art that expresses creativity.",
-    "Ballet is a classical dance form that requires discipline and skill.",
-    "Streaming platforms have changed the way we consume media.",
-    "The Earth orbits the sun in approximately 365 days.",
-    "Albert Einstein developed the theory of relativity.",
-    "Black holes are regions of space with extremely strong gravity.",
-    "DNA carries the genetic information of living organisms.",
-    "The Hubble Telescope has captured stunning images of distant galaxies.",
-    "Gravity is the force that keeps planets in orbit.",
-    "The periodic table organizes all known chemical elements.",
-    "Mars is often called the 'Red Planet' due to its color.",
-    "Photosynthesis is the process by which plants make their food.",
-    "The speed of light is approximately 299,792 kilometers per second.",
-    "Entrepreneurship involves taking risks to start a business.",
-    "Investing in stocks can yield high returns over time.",
-    "Marketing is essential for promoting products and services.",
-    "Small businesses are the backbone of many economies.",
-    "Cryptocurrencies like Bitcoin are decentralized digital currencies.",
-    "A good credit score is important for securing loans.",
-    "E-commerce has transformed the way people shop.",
-    "Networking is crucial for career advancement.",
-    "Inflation affects the purchasing power of money.",
-    "Sustainability is becoming a key focus for many companies.",
-    "Kindness is a universal language that everyone understands.",
-    "Time management is essential for achieving goals.",
-    "Laughter is often called the best medicine.",
-    "Volunteering can make a positive impact on your community.",
-    "Patience is a virtue that leads to better decision-making.",
-    "Honesty builds trust in relationships.",
-    "Curiosity drives innovation and discovery.",
-    "Gratitude can improve mental health and happiness.",
-    "Challenges help us grow and become stronger.",
-    "Life is a journey filled with ups and downs.",
-]
-
-def insert_all_documents(muopdb_client, collection_name, docs):
-    print("Inserting documents...")
-    id = 1
-    for sentence in docs:
-        result = genai.embed_content(
-                model="models/text-embedding-004",
-                content=sentence)
+import ollama
+import logging
+import h5py
+
+def insert_all_documents(muopdb_client: mp.IndexServerClient, collection_name, embeddings):
+    logging.info("Inserting documents...")
+    batch_size = 10000
+    total_embeddings = len(embeddings)
+    
+    for start_idx in range(0, total_embeddings, batch_size):
+        end_idx = min(start_idx + batch_size, total_embeddings)
+        batch_embeddings = embeddings[start_idx:end_idx]
+        batch_ids = list(range(start_idx + 1, end_idx + 1))
+        
+        # Flatten the batch embeddings into a single list of floats
+        flattened_vectors = [float(value) for embedding in batch_embeddings for value in embedding]
         muopdb_client.insert(
             collection_name=collection_name,
-            ids=[id],
-            vectors=result["embedding"]
+            ids=batch_ids,
+            vectors=flattened_vectors
         )
-        if id % 10 == 0:
-            print(f"Inserted document up to id {id}")
-        id += 1
+        
+        if end_idx % (batch_size * 10) == 0 or end_idx == total_embeddings:
+            logging.info(f"Inserted documents up to id {end_idx}")
+
+    logging.info("Start indexing documents...") 
     muopdb_client.flush(collection_name=collection_name)
+    logging.info("Indexing documents completed. Documents are ready to be queried.")
 
 # main function
 def main():
-    with open(os.path.expanduser('~/.secrets/gemini.config'), 'r') as f:
-        api_key = f.read().strip()
-    genai.configure(api_key=api_key)
-
-    query = "personal development"
-    print(f"Query: {query}")
-    query_vector = genai.embed_content(
-        model="models/text-embedding-004",
-        content=query)["embedding"]
-
-    # Before inserting documents, there shouldn't be any results
-    muopdb_client = mp.IndexServerClient()
-    search_response = muopdb_client.search(
-        index_name="test-collection-1",
-        vector=query_vector,
-        top_k=5,
-        ef_construction=50,
-        record_metrics=False
+    # Configure logging
+    logging.basicConfig(
+        level=logging.INFO,
+        format='%(asctime)s - %(levelname)s - %(message)s',
+        handlers=[
+            logging.StreamHandler()
+        ]
     )
-    print("Before inserting documents, number of results: ", len(search_response.ids))
+    
+    muopdb_client = mp.IndexServerClient()
+    logging.info("=========== Inserting documents ===========")
 
-    print("=========== Inserting documents ===========")
-    insert_all_documents(muopdb_client, "test-collection-1", SENTENCES)
-    print("=========== Inserted all documents ===========")
+    # Read the source embedding file
+    with h5py.File("/mnt/muopdb/raw/1m_embeddings.hdf5", "r") as f:
+        embeddings = f['embeddings'][:]
 
-    # After inserting documents, there should be results
-    print("Query: ", query)
-    search_response = muopdb_client.search(
-        index_name="test-collection-1",
-        vector=query_vector,
-        top_k=5,
-        ef_construction=50,
-        record_metrics=False
-    )
-    print("After inserting documents, number of results: ", len(search_response.ids))
-    for id in search_response.ids:
-        print(f"RESULT: {SENTENCES[id - 1]}")
+    # Insert the embeddings into MuopDB
+    insert_all_documents(muopdb_client, "test-collection-1", embeddings)
+    logging.info("=========== Inserted all documents ===========")
 
 
 if __name__ == "__main__":
diff --git a/py/demo_search.py b/py/demo_search.py
new file mode 100644
index 0000000..3287282
--- /dev/null
+++ b/py/demo_search.py
@@ -0,0 +1,30 @@
+import muopdb_client as mp
+import ollama
+import time
+
+if __name__ == "__main__":
+    # Example usage for IndexServer
+    muopdb_client = mp.IndexServerClient()
+    query = "personal career development"
+    query_vector = ollama.embeddings(model='nomic-embed-text', prompt=query)["embedding"]
+
+    # Read back the raw data to print the responses
+    with open("/mnt/muopdb/raw/1m_sentences.txt", "r") as f:
+        sentences = [line.strip() for line in f]
+
+    start = time.time()
+    search_response = muopdb_client.search(
+        index_name="test-collection-1",
+        vector=query_vector,
+        top_k=5,
+        ef_construction=50,
+        record_metrics=False
+    )
+    end = time.time()
+    print(f"Time taken for search: {end - start} seconds")
+
+    print(f"Number of results: {len(search_response.ids)}")
+    print("================")
+    for id in search_response.ids:
+        print(f"RESULT: {sentences[id - 1]}")
+    print("================")
diff --git a/py/embed_1m_sentences.py b/py/embed_1m_sentences.py
new file mode 100644
index 0000000..5679ba8
--- /dev/null
+++ b/py/embed_1m_sentences.py
@@ -0,0 +1,19 @@
+import ollama
+import time
+import sys
+
+if __name__ == "__main__":
+    # Read the first args
+    input_file = sys.argv[1]
+    with open(input_file, "r") as f:
+        sentences = f.readlines()
+
+    start = time.time()
+    output_file = f"{input_file}_embeddings.txt"
+    with open(output_file, "w") as f:
+        for sentence in sentences:
+            result = ollama.embeddings(model='nomic-embed-text', prompt=sentence)
+            f.write(f"{result['embedding']}\n")
+        
+    end = time.time()
+    print(f"Time taken: {end - start} seconds")
diff --git a/py/generate_1m_sentences.py b/py/generate_1m_sentences.py
new file mode 100644
index 0000000..69cd24d
--- /dev/null
+++ b/py/generate_1m_sentences.py
@@ -0,0 +1,9 @@
+import datasets
+
+if __name__ == "__main__":
+    ds = datasets.load_dataset("agentlans/high-quality-english-sentences")
+    train = ds["train"]
+    with open("/tmp/1m_sentences.txt", "w") as f:
+        for ln in train:
+            f.write(ln["text"] + "\n")
+    f.flush() 
diff --git a/py/muopdb_client.py b/py/muopdb_client.py
index 6d9fac7..5d7793a 100644
--- a/py/muopdb_client.py
+++ b/py/muopdb_client.py
@@ -4,7 +4,7 @@
 
 """
 Make sure to build the proto files first:
-pip install grpcio protobuf grpcio-tools
+pip install grpcio protobuf grpcio-tools ollama
 python3 -m grpc.tools.protoc -I=rs/proto/proto --python_out=py/ --grpc_python_out=py/ rs/proto/proto/muopdb.proto
 """
 
diff --git a/py/read_hdf5.py b/py/read_hdf5.py
new file mode 100644
index 0000000..e149b1d
--- /dev/null
+++ b/py/read_hdf5.py
@@ -0,0 +1,15 @@
+import h5py
+
+def read_first_and_last_five_points():
+    with h5py.File("/mnt/muopdb/raw/1m_embeddings.hdf5", "r") as f:
+        # Assuming the dataset is named 'embeddings'
+        dataset = f['embeddings']
+        first_five = dataset[:5]
+        last_five = dataset[-5:]
+        print("First 5 points:")
+        print(first_five)
+        print("\nLast 5 points:")
+        print(last_five)
+
+if __name__ == "__main__":
+    read_first_and_last_five_points()
\ No newline at end of file
diff --git a/rs/demo/Cargo.toml b/rs/demo/Cargo.toml
new file mode 100644
index 0000000..a59a81d
--- /dev/null
+++ b/rs/demo/Cargo.toml
@@ -0,0 +1,14 @@
+[package]
+name = "demo"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+tonic.workspace = true
+hdf5.workspace = true
+log.workspace = true
+env_logger.workspace = true
+anyhow.workspace = true
+proto.workspace = true
+tokio.workspace = true
+ndarray.workspace = true
diff --git a/rs/demo/src/main.rs b/rs/demo/src/main.rs
new file mode 100644
index 0000000..b526ff1
--- /dev/null
+++ b/rs/demo/src/main.rs
@@ -0,0 +1,74 @@
+use std::time::Instant;
+
+use anyhow::{Context, Result};
+use hdf5::File;
+use log::{LevelFilter, info};
+use proto::muopdb::index_server_client::IndexServerClient;
+use proto::muopdb::{FlushRequest, InsertRequest};
+use ndarray::s;
+
+#[tokio::main]
+async fn main() -> Result<()> {
+    // Configure logging
+    env_logger::Builder::new()
+        .filter_level(LevelFilter::Info)
+        .format_timestamp_millis()
+        .init();
+
+    let addr = String::from("http://127.0.0.1:9002");
+    // Create gRPC client
+    let mut client = IndexServerClient::connect(addr)
+        .await
+        .context("Failed to connect to IndexServer")?;
+
+    info!("=========== Inserting documents ===========");
+
+    // Read embeddings from HDF5 file
+    let file = File::open("/mnt/muopdb/raw/1m_embeddings.hdf5")?;
+    let dataset = file.dataset("embeddings")?;
+    let embeddings = dataset.read_2d::<f32>()?;
+
+    // Insert embeddings in batches into MuopDB
+    let batch_size = 100_000;
+    let total_embeddings = embeddings.nrows();
+    let mut start_idx = 0;
+
+    let mut start = Instant::now();
+    while start_idx < total_embeddings {
+        let end_idx = (start_idx + batch_size).min(total_embeddings);
+        let batch = &embeddings.slice(s![start_idx..end_idx, ..]);
+
+        let mut vectors = Vec::with_capacity(batch.len());
+        for row in batch.rows() {
+            vectors.extend(row.iter().map(|&v| v as f32));
+        }
+
+        // Generate IDs
+        let ids: Vec<u64> = (start_idx + 1..=end_idx).map(|i| i as u64).collect();
+
+        // Create and send insert request
+        let request = tonic::Request::new(InsertRequest {
+            collection_name: "test-collection-1".to_string(),
+            ids: ids.clone(),
+            vectors,
+        });
+
+        client.insert(request).await?;
+           start_idx = end_idx;
+    }
+
+    let mut duration = start.elapsed();
+    info!("Inserted all documents in {:?}", duration);
+
+    // Done inserting, now start indexing.
+    info!("Start indexing documents...");
+    start = Instant::now();
+    let request = tonic::Request::new(FlushRequest {
+        collection_name: "test-collection-1".to_string(),
+    });
+    client.flush(request).await?;
+    duration = start.elapsed();
+    info!("Indexing documents completed in {:?}", duration);
+
+    Ok(())
+}
diff --git a/rs/index_server/src/index_server.rs b/rs/index_server/src/index_server.rs
index b48bf06..15dedc2 100644
--- a/rs/index_server/src/index_server.rs
+++ b/rs/index_server/src/index_server.rs
@@ -3,7 +3,7 @@ use std::vec;
 
 use index::index::Searchable;
 use index::utils::SearchContext;
-use log::info;
+use log::{debug, info};
 use proto::muopdb::index_server_server::IndexServer;
 use proto::muopdb::{
     FlushRequest, FlushResponse, InsertRequest, InsertResponse, SearchRequest, SearchResponse,
@@ -87,6 +87,7 @@ impl IndexServer for IndexServerImpl {
         &self,
         request: tonic::Request<InsertRequest>,
     ) -> Result<tonic::Response<InsertResponse>, tonic::Status> {
+        let start = std::time::Instant::now();
         let req = request.into_inner();
         let collection_name = req.collection_name;
         let ids = req.ids;
@@ -117,6 +118,10 @@ impl IndexServer for IndexServerImpl {
                         collection.insert(*id, vector).unwrap()
                     });
 
+                // log the duration
+                let end = std::time::Instant::now();
+                let duration = end.duration_since(start);
+                debug!("Inserted {} vectors in {:?}", ids.len(), duration);
                 Ok(tonic::Response::new(InsertResponse { inserted_ids: ids }))
             }
             None => Err(tonic::Status::new(
@@ -130,6 +135,7 @@ impl IndexServer for IndexServerImpl {
         &self,
         request: tonic::Request<FlushRequest>,
     ) -> Result<tonic::Response<FlushResponse>, tonic::Status> {
+        let start = std::time::Instant::now();
         let req = request.into_inner();
         let collection_name = req.collection_name;
 
@@ -140,6 +146,10 @@ impl IndexServer for IndexServerImpl {
             .get_collection(&collection_name)
             .await;
 
+        let end = std::time::Instant::now();
+        let duration = end.duration_since(start);
+        debug!("Indexing collection {} in {:?}", collection_name, duration);
+
         match collection_opt {
             Some(collection) => {
                 collection.flush().unwrap();