diff --git a/.gitignore b/.gitignore
index 7ca1287..9c9eb81 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,5 +6,6 @@
 flamegraph.svg
 perf.data*
 .idea
+.DS_store
 .venv
 */__pycache__/
diff --git a/rs/utils/Cargo.toml b/rs/utils/Cargo.toml
index 0e60911..c749e37 100644
--- a/rs/utils/Cargo.toml
+++ b/rs/utils/Cargo.toml
@@ -24,6 +24,10 @@ harness = false
 name = "kmeans"
 harness = false
 
+[[bench]]
+name = "dot_product"
+harness = false
+
 [[bin]]
 name = "run_kmeans"
 path = "src/scripts/run_kmeans.rs"
diff --git a/rs/utils/benches/dot_product.rs b/rs/utils/benches/dot_product.rs
new file mode 100644
index 0000000..0c38634
--- /dev/null
+++ b/rs/utils/benches/dot_product.rs
@@ -0,0 +1,37 @@
+
+use criterion::{black_box, criterion_group, criterion_main, BenchmarkId, Criterion};
+use utils::test_utils::generate_random_vector;
+use utils::distance::dot_product::DotProductDistanceCalculator;
+use utils::DistanceCalculator;
+
+fn benches_dot_product(c: &mut Criterion) {
+    let mut group = c.benchmark_group("dot_product");
+
+    for size in [
+        8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096,
+        384,  // VECTOR_DIM_SENTENCE_TRANSFORMERS_MINI_LM
+        768,  // VECTOR_DIM_SENTENCE_TRANSFORMERS_MPNET
+        1536, // VECTOR_DIM_OPENAI_SMALL
+        3072, // VECTOR_DIM_OPENAI_LARGE
+    ]
+    .iter() {
+        let a = generate_random_vector(*size);
+        let b = generate_random_vector(*size);
+
+        group.bench_with_input(BenchmarkId::new("simd", size), &size, |bench, &_size| {
+            bench.iter(|| {
+                DotProductDistanceCalculator::calculate(black_box(&a), black_box(&b));
+            });
+        });
+
+        group.bench_with_input(BenchmarkId::new("scalar", size), &size, |bench, &_size| {
+            bench.iter(|| {
+                DotProductDistanceCalculator::calculate_scalar(black_box(&a), black_box(&b));
+            });
+        });
+    }
+    group.finish();
+}
+
+criterion_group!(benches, benches_dot_product);
+criterion_main!(benches);
diff --git a/rs/utils/src/distance/dot_product.rs b/rs/utils/src/distance/dot_product.rs
new file mode 100644
index 0000000..3c3a2bd
--- /dev/null
+++ b/rs/utils/src/distance/dot_product.rs
@@ -0,0 +1,96 @@
+use crate::DistanceCalculator;
+use std::{ops::AddAssign, simd::{num::SimdFloat, LaneCount, Simd, SupportedLaneCount}};
+
+pub struct DotProductDistanceCalculator {}
+
+impl DotProductDistanceCalculator {
+    pub fn calculate_scalar(a: &[f32], b: &[f32]) -> f32 {
+        let mut ret = 0.0;
+        for i in 0..a.len() {
+            ret += a[i] * b[i];
+        }
+        Self::neg_score(ret)
+    }
+
+    /* 
+     * In our code, the lower distance value, the greater similarity between two vectors.
+     * However, in dot product, two vector having the same direction 
+     * will yield the largest distance.
+     * Thus, we need to take negative value of the original dot product value. 
+     */
+    #[inline(always)]
+    pub fn neg_score(x: f32) -> f32 {
+        -x
+    }
+} 
+
+impl DistanceCalculator for DotProductDistanceCalculator {
+    #[inline(always)]
+    fn calculate(a: &[f32], b: &[f32]) -> f32 {
+        let mut res = 0.0;
+        let mut a_vec = a;
+        let mut b_vec = b;
+        
+        if a_vec.len() > 16 { 
+            let mut accumulator= Simd::<f32, 16>::splat(0.0);
+            Self::accumulate_lanes::<16>(a_vec, b_vec, &mut accumulator);
+            res += accumulator.reduce_sum();
+            a_vec = a_vec.chunks_exact(16).remainder();
+            b_vec = b_vec.chunks_exact(16).remainder();
+        }
+
+        if a_vec.len() > 8 { 
+            let mut accumulator= Simd::<f32, 8>::splat(0.0);
+            Self::accumulate_lanes::<8>(a_vec, b_vec, &mut accumulator);
+            res += accumulator.reduce_sum();
+            a_vec = a_vec.chunks_exact(8).remainder();
+            b_vec = b_vec.chunks_exact(8).remainder();
+        }
+
+        if a_vec.len() > 4 {
+            let mut accumulator= Simd::<f32, 4>::splat(0.0);
+            Self::accumulate_lanes::<4>(a_vec, b_vec, &mut accumulator);
+            res += accumulator.reduce_sum();
+            a_vec = a_vec.chunks_exact(4).remainder();
+            b_vec = b_vec.chunks_exact(4).remainder();
+        }
+
+        for i in 0..a_vec.len() {
+            res += a_vec[i] * b_vec[i];
+        }
+        Self::neg_score(res)
+    }
+
+    #[inline(always)]
+    fn accumulate_lanes<const LANES: usize>(
+        a: &[f32],
+        b: &[f32],
+        accumulator: &mut Simd<f32, LANES>,
+    ) where 
+        LaneCount<LANES>: SupportedLaneCount, 
+    {
+        a.chunks_exact(LANES)
+            .zip(b.chunks_exact(LANES))
+            .for_each(|(a_chunk, b_chunk)| {
+                let a_simd = Simd::<f32, LANES>::from_slice(a_chunk);
+                let b_simd = Simd::<f32, LANES>::from_slice(b_chunk);
+                accumulator.add_assign(a_simd * b_simd);
+            });
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::test_utils::generate_random_vector;
+    
+    #[test]
+    fn test_dot_product_distance_calculator() {
+        let a = generate_random_vector(128);
+        let b = generate_random_vector(128);
+        let eps = 1e-5;
+        let result = DotProductDistanceCalculator::calculate(&a, &b);
+        let expected = DotProductDistanceCalculator::calculate_scalar(&a, &b);
+        assert!((result - expected).abs() < eps);
+    }
+}
diff --git a/rs/utils/src/distance/mod.rs b/rs/utils/src/distance/mod.rs
index a9f3202..bc60355 100644
--- a/rs/utils/src/distance/mod.rs
+++ b/rs/utils/src/distance/mod.rs
@@ -1 +1,2 @@
 pub mod l2;
+pub mod dot_product;