diff --git a/.clang-tidy b/.clang-tidy
index 1da1dcfd48c2a28e2284deb46ce2b12cabfa9854..b2f304c5498279d31078465317912dc36f07ed40 100644
--- a/.clang-tidy
+++ b/.clang-tidy
@@ -4,33 +4,23 @@
 # SPDX-License-Identifier: Apache-2.0
 #
 ---
-Checks: '
--*,
-bugprone-*,
-cert-*,
-clang-analyzer-*,
-clang-diagnostic-*,
-cppcoreguidelines-*,
--cppcoreguidelines-owning-memory,
-google-*,
--google-readability-todo,
-llvm-*,
--llvm-include-order,
-misc-*,
-modernize-*,
--modernize-use-trailing-return-type,
-performance-*,
-readability-*,
--readability-identifier-length,
--readability-magic-numbers,
--readability-function-cognitive-complexity,
--cppcoreguidelines-pro-type-reinterpret-cast,
--cppcoreguidelines-avoid-magic-numbers,
--readability-simplify-boolean-expr,
--bugprone-easily-swappable-parameters,
--cppcoreguidelines-pro-bounds-pointer-arithmetic,
--performance-enum-size,
--llvm-else-after-return,
--readability-else-after-return,
-'
-...
+Checks: >
+  -*,
+  bugprone-*,
+  cert-*,
+  clang-analyzer-*,
+  clang-diagnostic-*,
+  cppcoreguidelines-*,
+  google-*,
+  llvm-*,
+  misc-*,
+  modernize-*,
+  performance-*,
+  readability-*,
+
+  -bugprone-easily-swappable-parameters,
+  -cppcoreguidelines-avoid-magic-numbers,
+  -llvm-header-guard,
+  -readability-function-cognitive-complexity,
+  -readability-identifier-length,
+  -readability-magic-numbers,
diff --git a/.editorconfig b/.editorconfig
index 8b3a3b254ce64477fd36659fddeba88c7fe15b97..7e95757f518756e8eca93e4f3001d967f1dd4d27 100644
--- a/.editorconfig
+++ b/.editorconfig
@@ -24,6 +24,10 @@ indent_size = unset
 [*.{json,yml,yaml}]
 indent_size = 2
 
+# YAML files as well.
+[.clang-{tidy,format}]
+indent_size = 2
+
 # Override settings.
 [LICENSES/*]
 indent_size = unset
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index e8b0b2f352bedf71aa3b39cd03046bcc6c847d60..9f8fff7257b7295ae8698e3ce282bb83a1314b02 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -142,13 +142,15 @@ test-examples:
     paths:
       - ${EXAMPLE}.log
 
-clang-tidy-checks:
+test-clang-tidy:
   extends:
     - .standard-rules
-  stage: build
+  stage: test
+  needs: []
   script:
-    - cmake -G Ninja -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release -DKLEIDIAI_BUILD_TESTS=ON -DKLEIDIAI_ENABLE_CLANG_TIDY=ON -S . -B ${CI_JOB_NAME_SLUG}
-    - cmake --build ${CI_JOB_NAME_SLUG}
+    - cmake -G Ninja -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release -DKLEIDIAI_BUILD_TESTS=ON -DKLEIDIAI_BUILD_BENCHMARK=ON -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -B build/${CI_JOB_NAME_SLUG}
+    # Only test the main library.
+    - run-clang-tidy -header-filter ".*" -warnings-as-errors "*" -p build/${CI_JOB_NAME_SLUG} kai
 
 pre-commit-hooks:
   variables:
diff --git a/kai/kai_common.h b/kai/kai_common.h
index 2703418542324f4c7a69635a88aae45a86d9d2df..944f4d069b61b31839871cd7be2bea47c9edf45d 100644
--- a/kai/kai_common.h
+++ b/kai/kai_common.h
@@ -92,7 +92,7 @@ inline static float kai_cast_f32_f16(uint16_t f16) {
 /// @return the f32 value
 inline static float kai_cast_f32_bf16(uint16_t bf16) {
     const uint32_t i32 = (bf16 << 16);
-    float f32;
+    float f32 = 0;
     memcpy(&f32, &i32, sizeof(i32));
     return f32;
 }
@@ -119,7 +119,7 @@ inline static uint16_t kai_cast_bf16_f32(float f32) {
 #if defined(__ARM_NEON)
 inline static uint16_t kai_cast_f16_f32(float f32) {
     uint16_t f16 = 0;
-    __fp16 tmp = f32;
+    __fp16 tmp = (__fp16)f32;
     memcpy(&f16, &tmp, sizeof(uint16_t));
     return f16;
 }
@@ -158,7 +158,10 @@ inline static uint64_t kai_get_sme_vector_length_u32(void) {
 ///
 /// @return the int8_t value with sign extended
 inline static int8_t kai_ext_sign_i8_i4(int8_t value) {
-    return (value ^ 0x8) - 8;
+    // Make sure value holds correct int4 value
+    KAI_ASSERT(value <= 0xF);
+
+    return (value ^ 0x8) - 8;  // NOLINT(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
 }
 
 #ifdef __cplusplus
diff --git a/kai/ukernels/matmul/matmul_clamp_f16_f16_f16p/kai_matmul_clamp_f16_f16_f16p16x1biasf16_6x16x8_neon_mla.c b/kai/ukernels/matmul/matmul_clamp_f16_f16_f16p/kai_matmul_clamp_f16_f16_f16p16x1biasf16_6x16x8_neon_mla.c
index 9f9d5a753284e88bf9bf4ac714d31fc9888ca20c..90335b31fa8660ca7c709ac86083dadee6e99fbf 100644
--- a/kai/ukernels/matmul/matmul_clamp_f16_f16_f16p/kai_matmul_clamp_f16_f16_f16p16x1biasf16_6x16x8_neon_mla.c
+++ b/kai/ukernels/matmul/matmul_clamp_f16_f16_f16p/kai_matmul_clamp_f16_f16_f16p16x1biasf16_6x16x8_neon_mla.c
@@ -11,7 +11,6 @@
 
 #include "kai_matmul_clamp_f16_f16_f16p16x1biasf16_6x16x8_neon_mla.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p_bf16p12x4b_8x12x4_neon_mmla.c b/kai/ukernels/matmul/matmul_clamp_f32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p_bf16p12x4b_8x12x4_neon_mmla.c
index 929e3753a90c40bb64072e34f2671d97b211bc53..4934496fea75e704bedae0b576ba50e81eaf6e47 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p_bf16p12x4b_8x12x4_neon_mmla.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p_bf16p12x4b_8x12x4_neon_mmla.c
@@ -13,6 +13,7 @@
 #include <arm_bf16.h>
 #include <arm_neon.h>
 #include <stddef.h>
+#include <stdint.h>
 
 #include "kai/kai_common.h"
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_f32_f32p/kai_matmul_clamp_f32_f32_f32p8x1biasf32_6x8x4_neon_mla.c b/kai/ukernels/matmul/matmul_clamp_f32_f32_f32p/kai_matmul_clamp_f32_f32_f32p8x1biasf32_6x8x4_neon_mla.c
index 62723018c615cc854912c4d450a96cb62a9f124e..fb380100906f6df59386a857d96b89a441ce7186 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_f32_f32p/kai_matmul_clamp_f32_f32_f32p8x1biasf32_6x8x4_neon_mla.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_f32_f32p/kai_matmul_clamp_f32_f32_f32p8x1biasf32_6x8x4_neon_mla.c
@@ -10,7 +10,6 @@
 
 #include "kai_matmul_clamp_f32_f32_f32p8x1biasf32_6x8x4_neon_mla.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_f32p_f32p/kai_matmul_clamp_f32_f32p2vlx1_f32p2vlx1biasf32_sme2_mopa.c b/kai/ukernels/matmul/matmul_clamp_f32_f32p_f32p/kai_matmul_clamp_f32_f32p2vlx1_f32p2vlx1biasf32_sme2_mopa.c
index 5611fad12f1802859f18d279469eb388b124c2a0..2836121f1216d8ab5dbad68c2f98cdf3db841045 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_f32p_f32p/kai_matmul_clamp_f32_f32p2vlx1_f32p2vlx1biasf32_sme2_mopa.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_f32p_f32p/kai_matmul_clamp_f32_f32p2vlx1_f32p2vlx1biasf32_sme2_mopa.c
@@ -76,8 +76,8 @@ void kai_run_matmul_clamp_f32_f32p2vlx1_f32p2vlx1biasf32_sme2_mopa(
         const void* B;
 
         void* C;
-        long ldcb;
-        long M, N, K;
+        uint64_t ldcb;
+        uint64_t M, N, K;
         float min;
         float max;
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
index d0a3f7de0fd65720ea5bc30488b185b948cd9488..532eb1ec33e9fe40fe2e7eff117ba37c0bcf6357 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
@@ -108,7 +108,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotp
 
 void kai_run_matmul_clamp_f32_qai8dxp1x8_qsi4c32p4x8_1x4x32_neon_dotprod(
     size_t m, size_t n, size_t k, size_t bl, const void* restrict lhs_packed, const void* restrict rhs_packed,
-    float* restrict dst, size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
+    size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT((bl % kai_kr) == 0);
     KAI_ASSERT((bl % kai_bl_multiple_of) == 0);
     KAI_ASSERT(dst_stride_col == sizeof(float));
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotprod.c
index ea48678ece9ce961c251bd6a73f5ace5f73e7a54..7f6f9e86b1a00aec4a2bdcd05b616d845ba68a8f 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotprod.c
@@ -108,7 +108,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotp
 
 void kai_run_matmul_clamp_f32_qai8dxp1x8_qsi4c32p8x8_1x8x32_neon_dotprod(
     size_t m, size_t n, size_t k, size_t bl, const void* restrict lhs_packed, const void* restrict rhs_packed,
-    float* restrict dst, size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
+    size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT((bl % kai_kr) == 0);
     KAI_ASSERT((bl % kai_bl_multiple_of) == 0);
     KAI_ASSERT(dst_stride_col == sizeof(float));
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8mm.c
index 92f8a56d63433d2a2147d1f8a205a754e1122bb5..7b4004427e638343ab4dc4724d66b4a1a865aab8 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8mm.c
@@ -107,7 +107,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8m
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_16x4x32_neon_i8mm(
-    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed, float* dst,
+    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT((bl % kai_kr) == 0);
     KAI_ASSERT((bl % kai_bl_multiple_of) == 0);
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
index 5e84e87099917affd92cb9df86fb16b87c3dc035..33ad7c3696f0a3d0683fec6b541f3bbc9f3102ac 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
@@ -107,7 +107,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4c32p4x8_8x4x32_neon_i8mm(
-    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed, float* dst,
+    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT((bl % kai_kr) == 0);
     KAI_ASSERT((bl % kai_bl_multiple_of) == 0);
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm.c
index f64249f137976a3e1b2501213a1c13d2dea321d5..62ffeb87f842e86e9ad3c97160e68e263df4747c 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4c32p/kai_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm.c
@@ -107,7 +107,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4c32p8x8_4x8x32_neon_i8mm(
-    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed, float* dst,
+    size_t m, size_t n, size_t k, size_t bl, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT((bl % kai_kr) == 0);
     KAI_ASSERT((bl % kai_bl_multiple_of) == 0);
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotprod.c
index e7cd60a7f0bf71398eba822f4562f3a85a06a203..e44a5fd945065b15ec71294a7bf1a973513ee9b6 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotprod.c
@@ -99,7 +99,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotpr
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp1x8_qsi4cxp4x8_1x4x32_neon_dotprod(
-    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed, float* restrict dst,
+    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed,
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotprod.c
index 0bde1f8beae3fde36c0679fdb31394f01c5ee2a4..54d63ae74d3fae56f11914c53bccadeea8bd8266 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotprod.c
@@ -99,7 +99,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotpr
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp1x8_qsi4cxp8x8_1x8x32_neon_dotprod(
-    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed, float* restrict dst,
+    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed,
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotprod.c
index 2ede94d6f677c2f05b76410a22920a9e51d81227..1f24fb465f89e526dca477bcd1acb55d92ddf1fb 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotprod.c
@@ -98,7 +98,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotp
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x4_16x4x32_neon_dotprod(
-    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed, float* restrict dst,
+    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed,
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm.c
index 4f8edd1e5ba8e3c0622f05cb8f2cc3eca69d0297..9b27c742b6eb4e542752db28bce2ba6b59cf27e0 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm.c
@@ -99,8 +99,9 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm(
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_4x4x32_neon_i8mm(
-    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed, float* dst, size_t dst_stride_row,
-    size_t dst_stride_col, float scalar_min, float scalar_max) {
+    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
+    size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
     if (m == 0) {
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm.c
index 7e40839e63d15b19a082461b033624ff7b22a5aa..7853163113d5ead9598cb9fb273bb0e2e485686a 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm.c
@@ -98,8 +98,9 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm(
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp4x8_8x4x32_neon_i8mm(
-    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed, float* dst, size_t dst_stride_row,
-    size_t dst_stride_col, float scalar_min, float scalar_max) {
+    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
+    size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
     if (m == 0) {
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotprod.c
index abe6f19fe2c613e88b63f73b96b4f7c8158f773c..99edd2c116d95fd55a3c23104516b83ce96e108e 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotprod.c
@@ -98,7 +98,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotpr
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x4_8x8x32_neon_dotprod(
-    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed, float* restrict dst,
+    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed,
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm.c
index fff2d5aa35bca78b0de4aecc0025abfaa0c59d1d..51585fe0b5522c5f2b5caa64938cb053df6a4dce 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm.c
@@ -98,7 +98,8 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm(
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_4x8x32_neon_i8mm(
-    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed, float* restrict dst,
+    size_t m, size_t n, size_t k, const void* restrict lhs_packed, const void* restrict rhs_packed,
+    float* restrict dst,  // NOLINT(readability-non-const-parameter)
     size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm.c
index e9f96a3ab2067fcb36da99bbba09ae66b69f30ac..b310bf748e42e1a4de205fbaa592be00daa25492 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qai8dxp_qsi4cxp/kai_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm.c
@@ -98,8 +98,9 @@ size_t kai_get_dst_size_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm(
 }
 
 void kai_run_matmul_clamp_f32_qai8dxp4x8_qsi4cxp8x8_8x8x32_neon_i8mm(
-    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed, float* dst, size_t dst_stride_row,
-    size_t dst_stride_col, float scalar_min, float scalar_max) {
+    size_t m, size_t n, size_t k, const void* lhs_packed, const void* rhs_packed,
+    float* dst,  // NOLINT(readability-non-const-parameter)
+    size_t dst_stride_row, size_t dst_stride_col, float scalar_min, float scalar_max) {
     KAI_ASSERT(dst_stride_col == sizeof(float));
 
     if (m == 0) {
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c b/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
index e41de7f6446c19f5253ee3707cd611c09d8103e8..37ee526cf0202f23342a686d5ebddcc88e10b93d 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1x8_qsi4c32p4x8_1x4x32_neon_dotprod.c
@@ -125,7 +125,7 @@ void kai_run_matmul_clamp_f32_qsi8d32p1x8_qsi4c32p4x8_1x4x32_neon_dotprod(
     const size_t num_rows = m;
     const size_t lhs_packed_stride = kai_lhs_packed_stride(k, bl);
 
-    const int8x16_t nibble_mask = vdupq_n_s8(0xF0);
+    const int8x16_t nibble_mask = vreinterpretq_s8_u8(vdupq_n_u8(0xF0));
 
     const uint8_t* lhs_ptr_start = lhs_packed;
 
diff --git a/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c b/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
index 78c9ec0b90884e18f16d9b589006b80b3e152c84..c3ce0882b35bed55d20c31b6660cf6c711adfd98 100644
--- a/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
+++ b/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm.c
@@ -10,6 +10,7 @@
 #include "kai_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm.h"
 
 #include <arm_neon.h>
+#include <stddef.h>
 #include <stdint.h>
 
 #include "kai/kai_common.h"
@@ -125,7 +126,7 @@ void kai_run_matmul_clamp_f32_qsi8d32p4x8_qsi4c32p4x8_8x4x32_neon_i8mm(
     const size_t num_cols = n;
     const size_t num_rows = m;
 
-    const int8x16_t nibble_mask = vdupq_n_s8(0xF0);
+    const int8x16_t nibble_mask = vreinterpretq_s8_u8(vdupq_n_u8(0xF0));
 
     const uint8_t* lhs_ptr_start = lhs_packed;
 
diff --git a/kai/ukernels/matmul/pack/kai_lhs_quant_pack_qsi8d32p_f32.c b/kai/ukernels/matmul/pack/kai_lhs_quant_pack_qsi8d32p_f32.c
index a391335df0224fca1d81b850703513d8fb09ca2a..e6c32004252a4fd1da62982897b922d151975bb5 100644
--- a/kai/ukernels/matmul/pack/kai_lhs_quant_pack_qsi8d32p_f32.c
+++ b/kai/ukernels/matmul/pack/kai_lhs_quant_pack_qsi8d32p_f32.c
@@ -91,20 +91,20 @@ void kai_run_lhs_quant_pack_qsi8d32p_f32(
 
             // Calculate scale and reciprocal
             const float scale = abs_max / ((1 << 7) - 1);
-            const float rep_scale = scale ? 1.0f / scale : 0.0f;
+            const float rep_scale = scale ? 1.0F / scale : 0.0F;
 
             // Quantize and pack the block
             for (size_t k_idx = 0; k_idx < bl; k_idx += k_block_len) {
-                for (size_t k_block_idx = 0; k_block_idx < (size_t)k_block_len; ++k_block_idx) {
+                for (size_t k_block_idx = 0; k_block_idx < k_block_len; ++k_block_idx) {
                     // Clamp at the last valid k-index
-                    const size_t k_idx_start = KAI_MIN((size_t)k_idx + k_block_idx, k - 1);
+                    const size_t k_idx_start = KAI_MIN(k_idx + k_block_idx, k - 1);
 
                     const float src0_0 = *(src_ptr + k_idx_start);
 
                     // Scale the values
                     int32_t v0_s32 = (int32_t)(roundf(src0_0 * rep_scale));
 
-                    *((int8_t*)(dst_ptr)) = (int8_t)v0_s32;
+                    *dst_ptr = (int8_t)v0_s32;
                     dst_ptr += sizeof(int8_t);
                 }
                 dst_ptr += (mr - 1) * k_block_len * sizeof(int8_t);
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme.c b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme.c
index 015e70bafe70a8fd24df9dbb7bd3e536efce3f04..5ce709fa256fba153637aaee3be76c991b64bb9f 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme.c
@@ -15,8 +15,8 @@
 
 static const size_t kai_nr = 2;
 static const size_t kai_kr = 1;
-static const size_t kai_num_bytes_data = sizeof(uint32_t);
-static const size_t kai_num_bytes_bias = sizeof(uint32_t);
+static const size_t kai_data_size_in_bytes = sizeof(uint32_t);
+static const size_t kai_bias_size_in_bytes = sizeof(uint32_t);
 
 size_t kai_get_n_step_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(void) {
     return kai_nr * kai_get_sme_vector_length_u32();
@@ -25,21 +25,21 @@ size_t kai_get_n_step_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(void) {
 size_t kai_get_rhs_offset_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(size_t n_idx) {
     KAI_ASSUME(n_idx % (kai_nr * kai_get_sme_vector_length_u32()) == 0);
 
-    return n_idx * kai_num_bytes_data;
+    return n_idx * kai_data_size_in_bytes;
 }
 
 size_t kai_get_bias_offset_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(size_t n_idx) {
-    return n_idx * kai_num_bytes_bias;
+    return n_idx * kai_bias_size_in_bytes;
 }
 
 size_t kai_get_rhs_packed_stride_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(size_t k) {
-    return kai_nr * kai_get_sme_vector_length_u32() * (kai_num_bytes_bias + k * kai_num_bytes_data);
+    return kai_nr * kai_get_sme_vector_length_u32() * (kai_bias_size_in_bytes + k * kai_data_size_in_bytes);
 }
 
 size_t kai_get_rhs_packed_offset_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(size_t n_idx, size_t k) {
     KAI_ASSUME(n_idx % (kai_nr * kai_get_sme_vector_length_u32()) == 0);
 
-    return n_idx * (kai_num_bytes_bias + k * kai_num_bytes_data);
+    return n_idx * (kai_bias_size_in_bytes + k * kai_data_size_in_bytes);
 }
 
 size_t kai_get_rhs_packed_size_rhs_pack_kxn_f32p2vlx1biasf32_f32_f32_sme(size_t n, size_t k) {
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32pb_f32_f32_16vlx1_sme.c b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32pb_f32_f32_16vlx1_sme.c
index 2e9685b8b2de7de625ebccd426a2cabc08cca63a..d03359ca54ef87985629df0e7af15d2090745a85 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32pb_f32_f32_16vlx1_sme.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_f32pb_f32_f32_16vlx1_sme.c
@@ -62,7 +62,7 @@ void kai_run_rhs_pack_kxn_f32pb_f32_f32_16vlx1_sme(
     const void* in = rhs;
     void* out = rhs_packed;
     const size_t in_stride = rhs_stride;
-    size_t out_stride = kai_nr * kai_get_sme_vector_length_u8() * (height + sizeof(uint32_t) / sizeof(uint32_t));
+    size_t out_stride = kai_nr * kai_get_sme_vector_length_u8() * (height + 1);
 
     __asm__ __volatile__(
         ".inst 0xd503477f  // SMSTART ZA\n"
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0.c b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0.c
index ece342ebd66aaaa2b1c7a323b7f15b41482c2497..ec191da45b304ca0fd4bf8077c959ea3992dd9db 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0.c
@@ -5,7 +5,6 @@
 //
 #include "kai_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -233,8 +232,8 @@ void kai_run_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0(
                     const uint8_t src_x0_lo = (byte0 & 0x0F);
                     const uint8_t src_x0_hi = (byte1 & 0x0F);
 
-                    sums[nr_idx] += ((int32_t)src_x0_lo - rhs_zero_point) * d;
-                    sums[nr_idx] += ((int32_t)src_x0_hi - rhs_zero_point) * d;
+                    sums[nr_idx] += (float)((int32_t)src_x0_lo - rhs_zero_point) * d;
+                    sums[nr_idx] += (float)((int32_t)src_x0_hi - rhs_zero_point) * d;
 
                     const uint8_t dst_qs0 = src_x0_lo | (src_x0_hi << 4);
 
@@ -243,8 +242,8 @@ void kai_run_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0(
                     const uint8_t src_x1_lo = (byte0 >> 4);
                     const uint8_t src_x1_hi = (byte1 >> 4);
 
-                    sums[nr_idx] += ((int32_t)src_x1_lo - rhs_zero_point) * d;
-                    sums[nr_idx] += ((int32_t)src_x1_hi - rhs_zero_point) * d;
+                    sums[nr_idx] += (float)((int32_t)src_x1_lo - rhs_zero_point) * d;
+                    sums[nr_idx] += (float)((int32_t)src_x1_hi - rhs_zero_point) * d;
 
                     const uint8_t dst_qs1 = src_x1_lo | (src_x1_hi << 4);
 
@@ -268,8 +267,5 @@ void kai_run_rhs_pack_kxn_qsi4c32p_qsu4c32s1s0(
                 ((float*)dst_row)[i] = bias[src_row_idx];
             }
         }
-
-        // Move the pointer after the biases
-        dst_row += kai_num_bytes_bias * nr;
     }
 }
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4cxp_qs4cxs1s0.c b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4cxp_qs4cxs1s0.c
index 491a7a34350209ade49b3c2da2551803a8d8880d..462218746eb8ee82736200714953e2233f69f463 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4cxp_qs4cxs1s0.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_qsi4cxp_qs4cxs1s0.c
@@ -5,7 +5,6 @@
 //
 #include "kai_rhs_pack_kxn_qsi4cxp_qs4cxs1s0.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -105,6 +104,8 @@ void kai_run_rhs_pack_kxn_qsi4cxp_qs4cxs1s0(
             const size_t src_addr_byte0 = (n0_valid_idx / 2) + k0_idx * rhs_stride;
             const size_t src_addr_byte1 = (n0_valid_idx / 2) + k1_idx * rhs_stride;
 
+            const size_t shift_right_x0 = (n0_idx % 2) * 4;
+
             if (params->rhs_zero_point == 8) {
                 uint8_t byte0 = rhs_zero_point | rhs_zero_point << 4;
                 uint8_t byte1 = rhs_zero_point | rhs_zero_point << 4;
@@ -135,7 +136,6 @@ void kai_run_rhs_pack_kxn_qsi4cxp_qs4cxs1s0(
                         src_x0_hi = (byte1 >> 4);
                     }
                 */
-                const size_t shift_right_x0 = (n0_idx % 2) * 4;
 
                 const uint8_t src_x0_lo = (byte0 >> shift_right_x0) & 0x0F;
                 const uint8_t src_x0_hi = (byte1 >> shift_right_x0) & 0x0F;
@@ -150,6 +150,7 @@ void kai_run_rhs_pack_kxn_qsi4cxp_qs4cxs1s0(
                 int8_t byte0 = 0;
                 int8_t byte1 = 0;
 
+                // NOLINTBEGIN(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
                 if (k0_idx < k) {
                     byte0 = rhs[src_addr_byte0];
                 }
@@ -161,12 +162,11 @@ void kai_run_rhs_pack_kxn_qsi4cxp_qs4cxs1s0(
                 // The logic behind the following operations where we extract the
                 // values from the bytes is same as unsigned
 
-                const size_t shift_right_x0 = (n0_idx % 2) * 4;
-
                 int8_t src_x0_lo = (byte0 >> shift_right_x0) & 0x0F;
                 int8_t src_x0_hi = (byte1 >> shift_right_x0) & 0x0F;
 
                 const int8_t dst_qs0 = src_x0_lo | (src_x0_hi << 4);
+                // NOLINTEND(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
 
                 *(int8_t*)dst_row = dst_qs0;
                 dst_row += sizeof(int8_t);
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0.c b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0.c
index f304324a34ea6ad0d72c94bdf5b16e9c0d0d18b7..269539671fae829b5cca198b405b850a5b9ad75f 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0.c
@@ -5,7 +5,6 @@
 //
 #include "kai_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -251,8 +250,8 @@ void kai_run_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0(
                         break;
                 }
 
-                sums[nr_idx] += ((int32_t)src_x0_lo - rhs_zero_point) * d;
-                sums[nr_idx] += ((int32_t)src_x0_hi - rhs_zero_point) * d;
+                sums[nr_idx] += (float)((int32_t)src_x0_lo - rhs_zero_point) * d;
+                sums[nr_idx] += (float)((int32_t)src_x0_hi - rhs_zero_point) * d;
 
                 const uint8_t dst_qs0 = src_x0_lo | (src_x0_hi << 4);
 
@@ -275,8 +274,5 @@ void kai_run_rhs_pack_nxk_qsi4c32p_qsu4c32s1s0(
                 ((float*)dst_row)[i] = bias[src_row_idx];
             }
         }
-
-        // Move the pointer after the biases
-        dst_row += kai_num_bytes_bias * nr;
     }
 }
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0.c b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0.c
index 475d781db28778030e44a4e3e807ecda1b0428c0..5d2f8454a2185e7899115f44f7f8b161bf4f4dda 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0.c
@@ -5,7 +5,6 @@
 //
 #include "kai_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -106,7 +105,7 @@ void kai_run_rhs_pack_nxk_qsi4c32pscalef16_qsu4c32s16s0(
     const size_t num_bytes_per_segment = kr / 2;
 
     for (size_t y = 0; y < n; y += nr) {
-        const uint8_t* src_row = (const uint8_t*)rhs + y * rhs_stride;
+        const uint8_t* src_row = rhs + y * rhs_stride;
         uint8_t* dst_row = (uint8_t*)rhs_packed + (y / nr) * rhs_packed_stride;
 
         for (size_t x = 0; x < num_blocks_per_row; ++x) {
diff --git a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4cxp_qs4cxs1s0.c b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4cxp_qs4cxs1s0.c
index 5a128f2ed33d6604c6e4aaf92ad61f99aded1d66..19887d096e9e0a9f3e8c9650b8c143dfa5b75bbe 100644
--- a/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4cxp_qs4cxs1s0.c
+++ b/kai/ukernels/matmul/pack/kai_rhs_pack_nxk_qsi4cxp_qs4cxs1s0.c
@@ -5,7 +5,6 @@
 //
 #include "kai_rhs_pack_nxk_qsi4cxp_qs4cxs1s0.h"
 
-#include <math.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -102,6 +101,9 @@ void kai_run_rhs_pack_nxk_qsi4cxp_qs4cxs1s0(
             const size_t src_addr_byte0 = (k0_idx / 2) + n0_valid_idx * rhs_stride;
             const size_t src_addr_byte1 = (k1_idx / 2) + n0_valid_idx * rhs_stride;
 
+            const size_t shift_right_x0 = (k0_idx % 2) * 4;
+            const size_t shift_right_x1 = (k1_idx % 2) * 4;
+
             if (params->rhs_zero_point == 8) {
                 uint8_t byte0 = rhs_zero_point | rhs_zero_point << 4;
                 uint8_t byte1 = rhs_zero_point | rhs_zero_point << 4;
@@ -132,9 +134,6 @@ void kai_run_rhs_pack_nxk_qsi4cxp_qs4cxs1s0(
                         src_x0_hi = (byte1 >> 4);
                     }
                 */
-                const size_t shift_right_x0 = (k0_idx % 2) * 4;
-                const size_t shift_right_x1 = (k1_idx % 2) * 4;
-
                 const uint8_t src_x0_lo = (byte0 >> shift_right_x0) & 0x0F;
                 const uint8_t src_x0_hi = (byte1 >> shift_right_x1) & 0x0F;
 
@@ -148,6 +147,7 @@ void kai_run_rhs_pack_nxk_qsi4cxp_qs4cxs1s0(
                 int8_t byte0 = 0;
                 int8_t byte1 = 0;
 
+                // NOLINTBEGIN(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
                 if (k0_idx < k) {
                     byte0 = rhs[src_addr_byte0];
                 }
@@ -159,13 +159,11 @@ void kai_run_rhs_pack_nxk_qsi4cxp_qs4cxs1s0(
                 // The logic behind the following operations where we extract the
                 // values from the bytes is same as unsigned
 
-                const size_t shift_right_x0 = (k0_idx % 2) * 4;
-                const size_t shift_right_x1 = (k1_idx % 2) * 4;
-
                 int8_t src_x0_lo = (byte0 >> shift_right_x0) & 0x0F;
                 int8_t src_x0_hi = (byte1 >> shift_right_x1) & 0x0F;
 
                 const int8_t dst_qs0 = src_x0_lo | (src_x0_hi << 4);
+                // NOLINTEND(bugprone-narrowing-conversions,cppcoreguidelines-narrowing-conversions)
 
                 *(int8_t*)dst_row = dst_qs0;
                 dst_row += sizeof(int8_t);