diff --git a/.jules/bolt.md b/.jules/bolt.md
index bf784525..1465d52e 100644
--- a/.jules/bolt.md
+++ b/.jules/bolt.md
@@ -25,3 +25,7 @@
 ## 2025-07-15 - Fast Bounding Box Pre-filter
 **Learning:** Calculating great circle distance (Haversine) for every issue against a target location is computationally expensive (O(N) with heavy math ops like sin, cos, atan2). In high-traffic aggregations, this can become a bottleneck.
 **Action:** Use a fast bounding box pre-filter (`get_bounding_box` with a 5% epsilon) to quickly discard issues that are definitely outside the search radius before running the expensive exact haversine distance calculation.
+
+## 2025-07-16 - N+1 Query in Escalation Loop
+**Learning:** During periodic evaluation of overdue grievances (`evaluate_and_escalate_grievances`), checking `grievance.jurisdiction.level` without eager loading triggers a separate DB query for every overdue grievance. This is a classic N+1 bottleneck when scanning a large number of open issues.
+**Action:** Always use `.options(joinedload(Model.relationship))` in queries that retrieve lists of items if those items' relationships will be accessed during iteration.
diff --git a/backend/escalation_engine.py b/backend/escalation_engine.py
index 67137b9b..2962811a 100644
--- a/backend/escalation_engine.py
+++ b/backend/escalation_engine.py
@@ -5,7 +5,7 @@
 
 import datetime
 from typing import List, Dict, Any, Optional
-from sqlalchemy.orm import Session
+from sqlalchemy.orm import Session, joinedload
 from sqlalchemy import and_, or_
 from backend.models import Grievance, Jurisdiction, EscalationAudit, GrievanceStatus, JurisdictionLevel, EscalationReason, SeverityLevel
 from backend.database import SessionLocal
@@ -150,7 +150,10 @@ def _get_grievances_for_evaluation(self, db: Session) -> List[Grievance]:
         now = datetime.datetime.now(datetime.timezone.utc)
 
         # Get grievances that are active and past SLA deadline
-        return db.query(Grievance).filter(
+        # PERFORMANCE OPTIMIZATION:
+        # Added .options(joinedload(Grievance.jurisdiction)) to resolve an N+1 query problem.
+        # This prevents a separate database query for each jurisdiction during the evaluation loop.
+        return db.query(Grievance).options(joinedload(Grievance.jurisdiction)).filter(
             and_(
                 Grievance.status.in_([GrievanceStatus.OPEN, GrievanceStatus.IN_PROGRESS, GrievanceStatus.ESCALATED]),
                 Grievance.sla_deadline < now