diff --git a/sqlite/util/graph_net_sample_groups_util.py b/sqlite/util/graph_net_sample_groups_util.py
new file mode 100644
index 000000000..39297fe48
--- /dev/null
+++ b/sqlite/util/graph_net_sample_groups_util.py
@@ -0,0 +1,49 @@
+# graph_net_sample_groups_util.py
+from typing import List, Set, Dict
+from collections import defaultdict
+
+from sqlite.orm_models import get_session, GraphNetSampleGroup
+
+
+def get_all_graph_net_sample_groups(
+    db_path: str,
+    group_types: List[str],
+    group_policies: List[str],
+    versions: List[str],
+) -> List[Set[str]]:
+    """
+    Get all graph_net sample groups from database.
+
+    Viba:
+        get_all_graph_net_sample_groups :=
+            list[set[$sample_uid str]]
+          <- $group_net_db_file_path str
+          <- $group_type list[str]
+          <- $group_policy list[str]
+          <- $version list[str]
+
+    Args:
+        db_path: Path to the SQLite database file.
+        group_types: List of group types to filter (e.g., ["shape_diversity", "dtype_diversity"]).
+        group_policies: List of group policies to filter (e.g., ["by_bucket"]).
+        versions: List of policy versions to filter (e.g., ["v0.1"]).
+
+    Returns:
+        List of sets, each set contains sample UIDs belonging to one group.
+    """
+    session = get_session(db_path)
+
+    query = session.query(GraphNetSampleGroup).filter(
+        GraphNetSampleGroup.deleted.is_(False),
+        GraphNetSampleGroup.group_type.in_(group_types),
+        GraphNetSampleGroup.group_policy.in_(group_policies),
+        GraphNetSampleGroup.policy_version.in_(versions),
+    )
+
+    groups_dict: Dict[str, List[str]] = defaultdict(list)
+    for row in query.all():
+        groups_dict[row.group_uid].append(row.sample_uid)
+
+    session.close()
+
+    return [set(uids) for uids in groups_dict.values()]
diff --git a/sqlite/util/graph_net_sample_util.py b/sqlite/util/graph_net_sample_util.py
new file mode 100644
index 000000000..e2d2c30ff
--- /dev/null
+++ b/sqlite/util/graph_net_sample_util.py
@@ -0,0 +1,127 @@
+# graph_net_sample_util.py
+import json
+from typing import Dict, List
+
+from sqlite.orm_models import get_session, GraphSample, SampleOpNameList
+
+
+class GraphNetSampleTypeGetter:
+    """
+    Get sample_type for a given sample_uid.
+
+    Viba:
+        GraphNetSampleTypeGetter :=
+            # __call__
+            $sample_type str
+          <- $sample_uid str
+            # __init__
+          <- $group_net_db_file_path str
+          <- $fetch_cache dict[$sample_uid str, $sample_type str]
+    """
+
+    def __init__(self, db_path: str):
+        self.db_path = db_path
+        self._cache: Dict[str, str] = {}
+
+    def __call__(self, sample_uid: str) -> str:
+        """Get sample_type for the given sample_uid."""
+        if sample_uid in self._cache:
+            return self._cache[sample_uid]
+
+        session = get_session(self.db_path)
+        sample = (
+            session.query(GraphSample).filter(GraphSample.uuid == sample_uid).first()
+        )
+        session.close()
+
+        sample_type = sample.sample_type if sample else ""
+        self._cache[sample_uid] = sample_type
+        return sample_type
+
+    def bulk_get(self, sample_uids: List[str]) -> Dict[str, str]:
+        """Bulk get sample_types for multiple sample UIDs."""
+        session = get_session(self.db_path)
+
+        samples = (
+            session.query(GraphSample).filter(GraphSample.uuid.in_(sample_uids)).all()
+        )
+
+        result = {}
+        for s in samples:
+            result[s.uuid] = s.sample_type
+            self._cache[s.uuid] = s.sample_type
+
+        for uid in sample_uids:
+            if uid not in result:
+                result[uid] = ""
+
+        session.close()
+        return result
+
+
+class GraphNetSampleOpSeqGetter:
+    """
+    Get op_seq for a given sample_uid.
+
+    Viba:
+        GraphNetSampleOpSeqGetter :=
+            # __call__
+            $sample_op_seq list[str]
+          <- $sample_uid str
+            # __init__
+          <- $group_net_db_file_path str
+          <- $fetch_cache dict[$sample_uid str, $sample_op_seq list[str]]
+    """
+
+    def __init__(self, db_path: str):
+        self.db_path = db_path
+        self._cache: Dict[str, List[str]] = {}
+
+    def __call__(self, sample_uid: str) -> List[str]:
+        """Get op_seq for the given sample_uid."""
+        if sample_uid in self._cache:
+            return self._cache[sample_uid]
+
+        session = get_session(self.db_path)
+        op_list = (
+            session.query(SampleOpNameList)
+            .filter(SampleOpNameList.sample_uuid == sample_uid)
+            .first()
+        )
+        session.close()
+
+        if op_list and op_list.op_names_json:
+            op_data = json.loads(op_list.op_names_json)
+            op_seq = [op["op_name"] for op in op_data]
+        else:
+            op_seq = []
+
+        self._cache[sample_uid] = op_seq
+        return op_seq
+
+    def bulk_get(self, sample_uids: List[str]) -> Dict[str, List[str]]:
+        """Bulk get op_seqs for multiple sample UIDs."""
+        session = get_session(self.db_path)
+
+        op_lists = (
+            session.query(SampleOpNameList)
+            .filter(SampleOpNameList.sample_uuid.in_(sample_uids))
+            .all()
+        )
+
+        result = {}
+        for op_list in op_lists:
+            if op_list.op_names_json:
+                op_data = json.loads(op_list.op_names_json)
+                op_seq = [op["op_name"] for op in op_data]
+            else:
+                op_seq = []
+            result[op_list.sample_uuid] = op_seq
+            self._cache[op_list.sample_uuid] = op_seq
+
+        for uid in sample_uids:
+            if uid not in result:
+                result[uid] = []
+
+        session.close()
+        return result