diff --git a/libkineto/CMakeLists.txt b/libkineto/CMakeLists.txt
index 6c5246481..78a51e963 100644
--- a/libkineto/CMakeLists.txt
+++ b/libkineto/CMakeLists.txt
@@ -30,8 +30,10 @@ set(KINETO_LIBRARY_TYPE "default" CACHE STRING
 set_property(CACHE KINETO_LIBRARY_TYPE PROPERTY STRINGS default shared)
 option(KINETO_BUILD_TESTS "Build kineto unit tests" ON)
 
+set(KINETO_BASE_DIR "${CMAKE_SOURCE_DIR}/../..")
 set(LIBKINETO_SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/src")
 set(LIBKINETO_INCLUDE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/include")
+set(LIBKINETO_UTILS_DIR "${CMAKE_CURRENT_SOURCE_DIR}/utils")
 set(LIBKINETO_BINARY_DIR ${CMAKE_CURRENT_BINARY_DIR})
 set(LIBKINETO_THIRDPARTY_DIR "${CMAKE_CURRENT_SOURCE_DIR}/third_party")
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
@@ -157,8 +159,10 @@ add_subdirectory("${IPCFABRIC_INCLUDE_DIR}")
 target_link_libraries(kineto_base PRIVATE dynolog_ipcfabric_lib)
 
 target_include_directories(kineto_base PUBLIC
+      $<BUILD_INTERFACE:${KINETO_BASE_DIR}>
       $<BUILD_INTERFACE:${LIBKINETO_INCLUDE_DIR}>
       $<BUILD_INTERFACE:${LIBKINETO_SOURCE_DIR}>
+      $<BUILD_INTERFACE:${LIBKINETO_UTILS_DIR}>
       $<BUILD_INTERFACE:${DYNOLOG_INCLUDE_DIR}>
       $<BUILD_INTERFACE:${FMT_INCLUDE_DIR}>
       $<BUILD_INTERFACE:${IPCFABRIC_INCLUDE_DIR}>
@@ -168,6 +172,7 @@ target_include_directories(kineto_base PUBLIC
       $<BUILD_INTERFACE:${ROCM_INCLUDE_DIRS}>)
 
 target_include_directories(kineto_api PUBLIC
+      $<BUILD_INTERFACE:${KINETO_BASE_DIR}>
       $<BUILD_INTERFACE:${FMT_INCLUDE_DIR}>
       $<BUILD_INTERFACE:${LIBKINETO_INCLUDE_DIR}>)
 
diff --git a/libkineto/include/GenericTraceActivity.h b/libkineto/include/GenericTraceActivity.h
index d840493f1..a4ee02b48 100644
--- a/libkineto/include/GenericTraceActivity.h
+++ b/libkineto/include/GenericTraceActivity.h
@@ -13,7 +13,7 @@
 #include <thread>
 #include <vector>
 
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 #include "ITraceActivity.h"
 #include "TraceSpan.h"
 
diff --git a/libkineto/include/libkineto.h b/libkineto/include/libkineto.h
index cd642e14c..9e02c500d 100644
--- a/libkineto/include/libkineto.h
+++ b/libkineto/include/libkineto.h
@@ -31,7 +31,7 @@
 #include "ILoggerObserver.h"
 #include "LoggingAPI.h"
 
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 extern "C" {
   void suppressLibkinetoLogMessages();
diff --git a/libkineto/ipcfabric/Utils.h b/libkineto/ipcfabric/Utils.h
index c32a8fcb4..f01479199 100644
--- a/libkineto/ipcfabric/Utils.h
+++ b/libkineto/ipcfabric/Utils.h
@@ -8,6 +8,9 @@
 
 #pragma once
 
+#include <sys/types.h>
+#include <cstdint>
+
 namespace dynolog {
 namespace ipcfabric {
 
diff --git a/libkineto/libkineto_defs.bzl b/libkineto/libkineto_defs.bzl
index 57e230dfc..b6b6a4f7b 100644
--- a/libkineto/libkineto_defs.bzl
+++ b/libkineto/libkineto_defs.bzl
@@ -5,14 +5,12 @@
 
 def get_libkineto_api_srcs():
     return [
-        "src/ThreadUtil.cpp",
         "src/libkineto_api.cpp",
+        "utils/ThreadUtil.cpp",
     ]
 
 def get_libkineto_cupti_srcs(with_api = True):
     return [
-        "src/CudaDeviceProperties.cpp",
-        "src/CudaUtil.cpp",
         "src/CuptiActivityApi.cpp",
         "src/CuptiActivityPlatform.cpp",
         "src/CuptiCallbackApi.cpp",
@@ -22,11 +20,12 @@ def get_libkineto_cupti_srcs(with_api = True):
         "src/CuptiRangeProfilerApi.cpp",
         "src/CuptiRangeProfilerConfig.cpp",
         "src/CuptiNvPerfMetric.cpp",
-        "src/Demangle.cpp",
         "src/EventProfiler.cpp",
         "src/EventProfilerController.cpp",
-        "src/WeakSymbols.cpp",
-        "src/cupti_strings.cpp",
+        "utils/CudaDeviceProperties.cpp",
+        "utils/CudaUtil.cpp",
+        "utils/WeakSymbols.cpp",
+        "utils/cupti_strings.cpp",
     ] + (get_libkineto_cpu_only_srcs(with_api))
 
 def get_libkineto_roctracer_srcs(with_api = True):
@@ -46,7 +45,6 @@ def get_libkineto_cpu_only_srcs(with_api = True):
         "src/ConfigLoader.cpp",
         "src/CuptiActivityApi.cpp",
         "src/DaemonConfigLoader.cpp",
-        "src/Demangle.cpp",
         "src/GenericTraceActivity.cpp",
         "src/ILoggerObserver.cpp",
         "src/IpcFabricConfigClient.cpp",
@@ -55,6 +53,7 @@ def get_libkineto_cpu_only_srcs(with_api = True):
         "src/init.cpp",
         "src/output_csv.cpp",
         "src/output_json.cpp",
+        "utils/Demangle.cpp",
     ] + (get_libkineto_api_srcs() if with_api else [])
 
 def get_libkineto_public_headers():
@@ -70,9 +69,10 @@ def get_libkineto_public_headers():
         "include/ILoggerObserver.h",
         "include/ITraceActivity.h",
         "include/TraceSpan.h",
-        "include/ThreadUtil.h",
         "include/libkineto.h",
         "include/time_since_epoch.h",
+        "utils/Demangle.h",
+        "utils/ThreadUtil.h",
     ]
 
 # kineto code should be updated to not have to
diff --git a/libkineto/src/ActivityProfilerController.cpp b/libkineto/src/ActivityProfilerController.cpp
index 74ce43370..9379f964b 100644
--- a/libkineto/src/ActivityProfilerController.cpp
+++ b/libkineto/src/ActivityProfilerController.cpp
@@ -18,7 +18,7 @@
 #ifdef HAS_ROCTRACER
 #include "RoctracerActivityApi.h"
 #endif
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 #include "output_json.h"
 #include "output_membuf.h"
 
diff --git a/libkineto/src/Config.cpp b/libkineto/src/Config.cpp
index b871f1f89..468912fda 100644
--- a/libkineto/src/Config.cpp
+++ b/libkineto/src/Config.cpp
@@ -23,7 +23,7 @@
 #include <time.h>
 
 #include "Logger.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 using namespace std::chrono;
 
diff --git a/libkineto/src/CuptiActivity.cpp b/libkineto/src/CuptiActivity.cpp
index 318d5d097..9b4482396 100644
--- a/libkineto/src/CuptiActivity.cpp
+++ b/libkineto/src/CuptiActivity.cpp
@@ -10,8 +10,8 @@
 
 #include <fmt/format.h>
 
-#include "CudaDeviceProperties.h"
-#include "Demangle.h"
+#include "kineto/libkineto/utils/CudaDeviceProperties.h"
+#include "kineto/libkineto/utils/Demangle.h"
 #include "output_base.h"
 
 namespace KINETO_NAMESPACE {
diff --git a/libkineto/src/CuptiActivity.h b/libkineto/src/CuptiActivity.h
index aef4ca83a..c90a61070 100644
--- a/libkineto/src/CuptiActivity.h
+++ b/libkineto/src/CuptiActivity.h
@@ -16,8 +16,8 @@
 #include "GenericTraceActivity.h"
 #include "CuptiActivityPlatform.h"
 #include "GenericTraceActivity.h"
-#include "ThreadUtil.h"
-#include "cupti_strings.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
+#include "kineto/libkineto/utils/cupti_strings.h"
 
 namespace libkineto {
   class ActivityLogger;
diff --git a/libkineto/src/CuptiActivityApi.cpp b/libkineto/src/CuptiActivityApi.cpp
index bdc9f7ffa..c5a8f7ef0 100644
--- a/libkineto/src/CuptiActivityApi.cpp
+++ b/libkineto/src/CuptiActivityApi.cpp
@@ -13,10 +13,10 @@
 #include <mutex>
 #include <thread>
 
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 #include "Logger.h"
 #include "Config.h"
-#include "CudaUtil.h"
+#include "kineto/libkineto/utils/CudaUtil.h"
 
 using namespace std::chrono;
 
diff --git a/libkineto/src/CuptiActivityProfiler.cpp b/libkineto/src/CuptiActivityProfiler.cpp
index 21c06a7ee..856d170eb 100644
--- a/libkineto/src/CuptiActivityProfiler.cpp
+++ b/libkineto/src/CuptiActivityProfiler.cpp
@@ -24,9 +24,9 @@
 #include <cupti.h>
 // TODO(T90238193)
 // @lint-ignore-every CLANGTIDY facebook-hte-RelativeInclude
-#include "cuda_call.h"
-#include "cupti_call.h"
-#include "CudaUtil.h"
+#include "kineto/libkineto/utils/cuda_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
+#include "kineto/libkineto/utils/CudaUtil.h"
 #endif
 
 #include "Config.h"
@@ -42,7 +42,7 @@
 #include "output_base.h"
 
 #include "Logger.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 using namespace std::chrono;
 using std::string;
diff --git a/libkineto/src/CuptiActivityProfiler.h b/libkineto/src/CuptiActivityProfiler.h
index 7fb6c71eb..9572ef999 100644
--- a/libkineto/src/CuptiActivityProfiler.h
+++ b/libkineto/src/CuptiActivityProfiler.h
@@ -30,7 +30,7 @@
 #include "CuptiActivity.h"
 #endif // HAS_CUPTI
 
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 #include "TraceSpan.h"
 #include "libkineto.h"
 #include "output_base.h"
diff --git a/libkineto/src/CuptiCallbackApi.cpp b/libkineto/src/CuptiCallbackApi.cpp
index 28fc27d44..678c5eab4 100644
--- a/libkineto/src/CuptiCallbackApi.cpp
+++ b/libkineto/src/CuptiCallbackApi.cpp
@@ -18,7 +18,7 @@
 #include <shared_mutex>
 
 #ifdef HAS_CUPTI
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 #endif
 #include "Logger.h"
 
diff --git a/libkineto/src/CuptiEventApi.cpp b/libkineto/src/CuptiEventApi.cpp
index fbae683fe..f90e7465d 100644
--- a/libkineto/src/CuptiEventApi.cpp
+++ b/libkineto/src/CuptiEventApi.cpp
@@ -11,7 +11,7 @@
 #include <chrono>
 
 #include "Logger.h"
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 
 using std::vector;
 
diff --git a/libkineto/src/CuptiMetricApi.cpp b/libkineto/src/CuptiMetricApi.cpp
index a9742e4c5..b9ca23821 100644
--- a/libkineto/src/CuptiMetricApi.cpp
+++ b/libkineto/src/CuptiMetricApi.cpp
@@ -11,7 +11,7 @@
 #include <chrono>
 
 #include "Logger.h"
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 
 using std::vector;
 
diff --git a/libkineto/src/CuptiNvPerfMetric.cpp b/libkineto/src/CuptiNvPerfMetric.cpp
index af6330f43..3e2a7b762 100644
--- a/libkineto/src/CuptiNvPerfMetric.cpp
+++ b/libkineto/src/CuptiNvPerfMetric.cpp
@@ -17,7 +17,7 @@
 
 // TODO(T90238193)
 // @lint-ignore-every CLANGTIDY facebook-hte-RelativeInclude
-#include "ScopeExit.h"
+#include "kineto/libkineto/utils/ScopeExit.h"
 #include "CuptiNvPerfMetric.h"
 #include "Logger.h"
 
diff --git a/libkineto/src/CuptiRangeProfiler.cpp b/libkineto/src/CuptiRangeProfiler.cpp
index c26b660fd..273b75a68 100644
--- a/libkineto/src/CuptiRangeProfiler.cpp
+++ b/libkineto/src/CuptiRangeProfiler.cpp
@@ -19,7 +19,7 @@
 #include "output_base.h"
 #include "CuptiRangeProfiler.h"
 #include "CuptiRangeProfilerConfig.h"
-#include "Demangle.h"
+#include "kineto/libkineto/utils/Demangle.h"
 
 namespace KINETO_NAMESPACE {
 
diff --git a/libkineto/src/CuptiRangeProfilerApi.cpp b/libkineto/src/CuptiRangeProfilerApi.cpp
index 4b3963b08..b66254105 100644
--- a/libkineto/src/CuptiRangeProfilerApi.cpp
+++ b/libkineto/src/CuptiRangeProfilerApi.cpp
@@ -16,12 +16,12 @@
 #include <unordered_map>
 
 #ifdef HAS_CUPTI
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 #endif
 
 #include "time_since_epoch.h"
 #include "Logger.h"
-#include "Demangle.h"
+#include "kineto/libkineto/utils/Demangle.h"
 
 // TODO(T90238193)
 // @lint-ignore-every CLANGTIDY facebook-hte-RelativeInclude
@@ -30,7 +30,7 @@
 #if HAS_CUPTI_RANGE_PROFILER
 #include <cupti.h>
 #include <nvperf_host.h>
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 #endif // HAS_CUPTI_RANGE_PROFILER
 
 namespace KINETO_NAMESPACE {
diff --git a/libkineto/src/EventProfilerController.cpp b/libkineto/src/EventProfilerController.cpp
index 55bf83033..a34beaa5d 100644
--- a/libkineto/src/EventProfilerController.cpp
+++ b/libkineto/src/EventProfilerController.cpp
@@ -20,7 +20,7 @@
 #include "output_csv.h"
 
 #include "Logger.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 using namespace std::chrono;
 using std::unique_ptr;
diff --git a/libkineto/src/IpcFabricConfigClient.cpp b/libkineto/src/IpcFabricConfigClient.cpp
index fc2970ce7..e39ad6413 100644
--- a/libkineto/src/IpcFabricConfigClient.cpp
+++ b/libkineto/src/IpcFabricConfigClient.cpp
@@ -17,7 +17,7 @@
 
 // TODO(T90238193)
 // @lint-ignore-every CLANGTIDY facebook-hte-RelativeInclude
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 namespace KINETO_NAMESPACE {
 
diff --git a/libkineto/src/Logger.cpp b/libkineto/src/Logger.cpp
index d4886506f..3e775944d 100644
--- a/libkineto/src/Logger.cpp
+++ b/libkineto/src/Logger.cpp
@@ -22,7 +22,7 @@
 #include <fmt/chrono.h>
 #include <fmt/format.h>
 
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 namespace KINETO_NAMESPACE {
 
diff --git a/libkineto/src/RoctracerActivityApi.cpp b/libkineto/src/RoctracerActivityApi.cpp
index 9fdfc5e45..6f51c705a 100644
--- a/libkineto/src/RoctracerActivityApi.cpp
+++ b/libkineto/src/RoctracerActivityApi.cpp
@@ -13,9 +13,9 @@
 
 #include "RoctracerActivityApi.h"
 #include "RoctracerLogger.h"
-#include "Demangle.h"
+#include "kineto/libkineto/utils/Demangle.h"
 #include "output_base.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 typedef uint64_t timestamp_t;
 
diff --git a/libkineto/src/RoctracerLogger.cpp b/libkineto/src/RoctracerLogger.cpp
index 4e781c451..f2d5dc707 100644
--- a/libkineto/src/RoctracerLogger.cpp
+++ b/libkineto/src/RoctracerLogger.cpp
@@ -12,7 +12,7 @@
 #include <chrono>
 #include <time.h>
 
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 typedef uint64_t timestamp_t;
 
@@ -360,4 +360,3 @@ bool ApiIdList::contains(uint32_t apiId)
 {
   return (filter_.find(apiId) != filter_.end()) ? !invert_ : invert_;  // XOR
 }
-
diff --git a/libkineto/src/init.cpp b/libkineto/src/init.cpp
index 60fc0fd4d..9f35352b1 100644
--- a/libkineto/src/init.cpp
+++ b/libkineto/src/init.cpp
@@ -20,7 +20,7 @@
 #include "CuptiRangeProfiler.h"
 #include "EventProfilerController.h"
 #endif
-#include "cupti_call.h"
+#include "kineto/libkineto/utils/cupti_call.h"
 #include "libkineto.h"
 
 #include "Logger.h"
diff --git a/libkineto/src/libkineto_api.cpp b/libkineto/src/libkineto_api.cpp
index 8f92fb97e..8be5783fa 100644
--- a/libkineto/src/libkineto_api.cpp
+++ b/libkineto/src/libkineto_api.cpp
@@ -9,7 +9,7 @@
 #include "libkineto.h"
 
 #include "ConfigLoader.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 
 namespace libkineto {
 
diff --git a/libkineto/src/output_base.h b/libkineto/src/output_base.h
index 2438fe400..b13062f15 100644
--- a/libkineto/src/output_base.h
+++ b/libkineto/src/output_base.h
@@ -16,7 +16,7 @@
 
 #include "ActivityBuffers.h"
 #include "GenericTraceActivity.h"
-#include "ThreadUtil.h"
+#include "kineto/libkineto/utils/ThreadUtil.h"
 #include "TraceSpan.h"
 
 namespace KINETO_NAMESPACE {
diff --git a/libkineto/src/output_json.cpp b/libkineto/src/output_json.cpp
index 15735aa85..79bdc05cb 100644
--- a/libkineto/src/output_json.cpp
+++ b/libkineto/src/output_json.cpp
@@ -15,7 +15,7 @@
 
 #include "Config.h"
 #ifdef HAS_CUPTI
-#include "CudaDeviceProperties.h"
+#include "kineto/libkineto/utils/CudaDeviceProperties.h"
 #endif // HAS_CUPTI
 #include "TraceSpan.h"
 
diff --git a/libkineto/test/CuptiStringsTest.cpp b/libkineto/test/CuptiStringsTest.cpp
index afb3fd0c3..62ba2749e 100644
--- a/libkineto/test/CuptiStringsTest.cpp
+++ b/libkineto/test/CuptiStringsTest.cpp
@@ -8,7 +8,7 @@
 
 #include <gtest/gtest.h>
 
-#include "src/cupti_strings.h"
+#include "utils/cupti_strings.h"
 
 using namespace KINETO_NAMESPACE;
 
diff --git a/libkineto/test/PidInfoTest.cpp b/libkineto/test/PidInfoTest.cpp
index cb4069240..7ad4ee9f9 100644
--- a/libkineto/test/PidInfoTest.cpp
+++ b/libkineto/test/PidInfoTest.cpp
@@ -6,7 +6,7 @@
  * LICENSE file in the root directory of this source tree.
  */
 
-#include "include/ThreadUtil.h"
+#include "utils/ThreadUtil.h"
 
 #include <atomic>
 #include <thread>
diff --git a/libkineto/src/CudaDeviceProperties.cpp b/libkineto/utils/CudaDeviceProperties.cpp
similarity index 100%
rename from libkineto/src/CudaDeviceProperties.cpp
rename to libkineto/utils/CudaDeviceProperties.cpp
diff --git a/libkineto/src/CudaDeviceProperties.h b/libkineto/utils/CudaDeviceProperties.h
similarity index 100%
rename from libkineto/src/CudaDeviceProperties.h
rename to libkineto/utils/CudaDeviceProperties.h
diff --git a/libkineto/src/CudaUtil.cpp b/libkineto/utils/CudaUtil.cpp
similarity index 100%
rename from libkineto/src/CudaUtil.cpp
rename to libkineto/utils/CudaUtil.cpp
diff --git a/libkineto/src/CudaUtil.h b/libkineto/utils/CudaUtil.h
similarity index 100%
rename from libkineto/src/CudaUtil.h
rename to libkineto/utils/CudaUtil.h
diff --git a/libkineto/src/Demangle.cpp b/libkineto/utils/Demangle.cpp
similarity index 100%
rename from libkineto/src/Demangle.cpp
rename to libkineto/utils/Demangle.cpp
diff --git a/libkineto/src/Demangle.h b/libkineto/utils/Demangle.h
similarity index 100%
rename from libkineto/src/Demangle.h
rename to libkineto/utils/Demangle.h
diff --git a/libkineto/src/ScopeExit.h b/libkineto/utils/ScopeExit.h
similarity index 100%
rename from libkineto/src/ScopeExit.h
rename to libkineto/utils/ScopeExit.h
diff --git a/libkineto/src/ThreadUtil.cpp b/libkineto/utils/ThreadUtil.cpp
similarity index 100%
rename from libkineto/src/ThreadUtil.cpp
rename to libkineto/utils/ThreadUtil.cpp
diff --git a/libkineto/include/ThreadUtil.h b/libkineto/utils/ThreadUtil.h
similarity index 100%
rename from libkineto/include/ThreadUtil.h
rename to libkineto/utils/ThreadUtil.h
diff --git a/libkineto/src/WeakSymbols.cpp b/libkineto/utils/WeakSymbols.cpp
similarity index 100%
rename from libkineto/src/WeakSymbols.cpp
rename to libkineto/utils/WeakSymbols.cpp
diff --git a/libkineto/src/cuda_call.h b/libkineto/utils/cuda_call.h
similarity index 100%
rename from libkineto/src/cuda_call.h
rename to libkineto/utils/cuda_call.h
diff --git a/libkineto/src/cupti_call.h b/libkineto/utils/cupti_call.h
similarity index 100%
rename from libkineto/src/cupti_call.h
rename to libkineto/utils/cupti_call.h
diff --git a/libkineto/src/cupti_strings.cpp b/libkineto/utils/cupti_strings.cpp
similarity index 100%
rename from libkineto/src/cupti_strings.cpp
rename to libkineto/utils/cupti_strings.cpp
diff --git a/libkineto/src/cupti_strings.h b/libkineto/utils/cupti_strings.h
similarity index 100%
rename from libkineto/src/cupti_strings.h
rename to libkineto/utils/cupti_strings.h