diff --git a/gluten-substrait/src/main/scala/org/apache/gluten/utils/PullOutProjectHelper.scala b/gluten-substrait/src/main/scala/org/apache/gluten/utils/PullOutProjectHelper.scala
index e4fc11441031..4bb09bfcab6a 100644
--- a/gluten-substrait/src/main/scala/org/apache/gluten/utils/PullOutProjectHelper.scala
+++ b/gluten-substrait/src/main/scala/org/apache/gluten/utils/PullOutProjectHelper.scala
@@ -20,7 +20,7 @@ import org.apache.gluten.backendsapi.BackendsApiManager
 import org.apache.gluten.exception.{GlutenException, GlutenNotSupportException}
 
 import org.apache.spark.sql.catalyst.expressions._
-import org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression, AggregateFunction}
+import org.apache.spark.sql.catalyst.expressions.aggregate.{AggregateExpression, AggregateFunction, Complete, Partial}
 import org.apache.spark.sql.execution.aggregate._
 import org.apache.spark.sql.execution.window.WindowExec
 import org.apache.spark.sql.types.{ByteType, DateType, IntegerType, LongType, ShortType}
@@ -137,16 +137,20 @@ trait PullOutProjectHelper {
   protected def rewriteAggregateExpression(
       ae: AggregateExpression,
       expressionMap: mutable.HashMap[Expression, NamedExpression]): AggregateExpression = {
-    val newAggFuncChildren = ae.aggregateFunction.children.map {
-      case literal: Literal => literal
-      case other => replaceExpressionWithAttribute(other, expressionMap)
+    ae.mode match {
+      case Partial | Complete =>
+        val newAggFuncChildren = ae.aggregateFunction.children.map {
+          case literal: Literal => literal
+          case other => replaceExpressionWithAttribute(other, expressionMap)
+        }
+        val newAggFunc = ae.aggregateFunction
+          .withNewChildren(newAggFuncChildren)
+          .asInstanceOf[AggregateFunction]
+        val newFilter =
+          ae.filter.map(replaceExpressionWithAttribute(_, expressionMap))
+        ae.copy(aggregateFunction = newAggFunc, filter = newFilter)
+      case _ => ae
     }
-    val newAggFunc = ae.aggregateFunction
-      .withNewChildren(newAggFuncChildren)
-      .asInstanceOf[AggregateFunction]
-    val newFilter =
-      ae.filter.map(replaceExpressionWithAttribute(_, expressionMap))
-    ae.copy(aggregateFunction = newAggFunc, filter = newFilter)
   }
 
   private def needPreComputeRangeFrameBoundary(bound: Expression): Boolean = {
diff --git a/gluten-substrait/src/test/scala/org/apache/spark/sql/GlutenQueryTest.scala b/gluten-substrait/src/test/scala/org/apache/spark/sql/GlutenQueryTest.scala
index 1d4a63bdbc28..8852fdb137b7 100644
--- a/gluten-substrait/src/test/scala/org/apache/spark/sql/GlutenQueryTest.scala
+++ b/gluten-substrait/src/test/scala/org/apache/spark/sql/GlutenQueryTest.scala
@@ -397,6 +397,28 @@ abstract class GlutenQueryTest extends PlanTest {
     val executedPlan = getExecutedPlan(df)
     assert(executedPlan.exists(plan => tag.runtimeClass.isInstance(plan)))
   }
+
+  /**
+   * Check whether the executed plan of a dataframe contains expected number of expected plans.
+   *
+   * @param df:
+   *   the input dataframe.
+   * @param count:
+   *   expected number of expected plan.
+   * @param tag:
+   *   class of the expected plan.
+   * @tparam T:
+   *   type of the expected plan.
+   */
+  def checkGlutenOperatorCount[T <: GlutenPlan](df: DataFrame, count: Int)(implicit
+      tag: ClassTag[T]): Unit = {
+    val executedPlan = getExecutedPlan(df)
+    assert(
+      executedPlan.count(plan => tag.runtimeClass.isInstance(plan)) == count,
+      s"Expect $count ${tag.runtimeClass.getSimpleName} " +
+        s"in executedPlan:\n ${executedPlan.last}"
+    )
+  }
 }
 
 object GlutenQueryTest extends Assertions {
diff --git a/gluten-ut/test/src/test/scala/org/apache/gluten/extension/GlutenExtensionRewriteRuleSuite.scala b/gluten-ut/test/src/test/scala/org/apache/gluten/extension/GlutenExtensionRewriteRuleSuite.scala
index a29550815041..837d37236b65 100644
--- a/gluten-ut/test/src/test/scala/org/apache/gluten/extension/GlutenExtensionRewriteRuleSuite.scala
+++ b/gluten-ut/test/src/test/scala/org/apache/gluten/extension/GlutenExtensionRewriteRuleSuite.scala
@@ -16,7 +16,7 @@
  */
 package org.apache.gluten.extension
 
-import org.apache.gluten.execution.{ProjectExecTransformer, WholeStageTransformerSuite}
+import org.apache.gluten.execution.{HashAggregateExecBaseTransformer, ProjectExecTransformer, WholeStageTransformerSuite}
 import org.apache.gluten.utils.BackendTestUtils
 
 import org.apache.spark.SparkConf
@@ -62,4 +62,31 @@ class GlutenExtensionRewriteRuleSuite extends WholeStageTransformerSuite {
       }
     )
   }
+
+  test("GLUTEN-9279 - Not Pull out expression to avoid invalid reference binding") {
+    withTable("t") {
+      sql("CREATE TABLE t(f1 String, f2 String, f3 String, f4 String) USING PARQUET")
+      sql("INSERT INTO t values ('1', '2', '3', '4'), ('11' ,'22', '33', '4')")
+      var expectedProjectCount = 3
+      var noFallback = false
+      if (BackendTestUtils.isCHBackendLoaded()) {
+        // The `RewriteMultiChildrenCount` rule in the Velox-backend is the root cause of the
+        // additional ProjectExecTransformer, which leads to the invalid reference binding issue.
+        // We still conduct tests on the CH-backend here to ensure that the introduced modification
+        // in `PullOutPreProject` has no side effect on the CH-backend.
+        expectedProjectCount = 2
+        noFallback = true
+      }
+      runQueryAndCompare(
+        """
+          |SELECT SUM(f1) / COUNT(DISTINCT f2, f3) FROM t GROUP BY f4;
+          |""".stripMargin,
+        noFallBack = noFallback
+      )(
+        df => {
+          checkGlutenOperatorCount[ProjectExecTransformer](df, expectedProjectCount)
+          checkGlutenOperatorCount[HashAggregateExecBaseTransformer](df, 4)
+        })
+    }
+  }
 }