diff --git a/app/services/db_aggregator.py b/app/services/db_aggregator.py
new file mode 100644
index 0000000..830ec3f
--- /dev/null
+++ b/app/services/db_aggregator.py
@@ -0,0 +1,610 @@
+import json
+from datetime import datetime, timedelta
+from typing import List, Dict, Any, Optional, Tuple, Union
+from sqlalchemy.orm import Session
+from sqlalchemy import and_, or_, func, desc, text, case, cast, String
+from sqlalchemy.sql import select
+
+from app.models.transaction import Transaction
+from app.models.screening import AggregateCache
+
+class DatabaseAggregator:
+    """
+    High-performance database-level aggregation service.
+    Uses native SQL aggregation functions for optimal performance.
+    """
+    
+    def __init__(self, db: Session):
+        self.db = db
+    
+    def aggregate_transactions(
+        self,
+        aggregate_function: str,
+        field: str,
+        group_by: Optional[List[str]] = None,
+        filters: Optional[Dict[str, Any]] = None,
+        time_window: Optional[str] = None,
+        date_from: Optional[datetime] = None,
+        date_to: Optional[datetime] = None,
+        use_cache: bool = True
+    ) -> Dict[str, Any]:
+        """
+        Perform database-level aggregation on transactions.
+        
+        Args:
+            aggregate_function: sum, count, avg, max, min, count_distinct
+            field: Field to aggregate on (* for count)
+            group_by: List of fields to group by
+            filters: Filter conditions
+            time_window: Time window (e.g., "24h", "7d")
+            date_from: Start date filter
+            date_to: End date filter
+            use_cache: Whether to use caching
+            
+        Returns:
+            Dictionary with aggregation results
+        """
+        # Generate cache key
+        cache_key = self._generate_cache_key(
+            aggregate_function, field, group_by, filters, time_window, date_from, date_to
+        )
+        
+        # Check cache if enabled
+        if use_cache:
+            cached_result = self._get_cached_result(cache_key)
+            if cached_result:
+                return cached_result
+        
+        # Build query
+        query = self._build_aggregation_query(
+            aggregate_function, field, group_by, filters, time_window, date_from, date_to
+        )
+        
+        # Execute query
+        start_time = datetime.utcnow()
+        results = query.all()
+        execution_time = (datetime.utcnow() - start_time).total_seconds() * 1000
+        
+        # Format results
+        formatted_results = self._format_aggregation_results(
+            results, aggregate_function, field, group_by
+        )
+        
+        # Prepare response
+        response = {
+            "results": formatted_results,
+            "metadata": {
+                "aggregate_function": aggregate_function,
+                "field": field,
+                "group_by": group_by,
+                "filters": filters,
+                "time_window": time_window,
+                "execution_time_ms": execution_time,
+                "total_groups": len(formatted_results) if group_by else 1,
+                "cache_hit": False,
+                "computed_at": start_time.isoformat()
+            }
+        }
+        
+        # Cache result if enabled
+        if use_cache:
+            self._cache_result(cache_key, response, time_window)
+        
+        return response
+    
+    def aggregate_by_user(
+        self,
+        aggregate_function: str,
+        field: str,
+        user_id: Optional[str] = None,
+        time_window: Optional[str] = None,
+        filters: Optional[Dict[str, Any]] = None
+    ) -> Dict[str, Any]:
+        """
+        Aggregate transactions by user.
+        """
+        group_by = ["user_id"]
+        if user_id:
+            filters = filters or {}
+            filters["user_id"] = user_id
+            group_by = None  # Don't group if filtering by specific user
+        
+        return self.aggregate_transactions(
+            aggregate_function=aggregate_function,
+            field=field,
+            group_by=group_by,
+            filters=filters,
+            time_window=time_window
+        )
+    
+    def aggregate_by_account(
+        self,
+        aggregate_function: str,
+        field: str,
+        account_id: Optional[str] = None,
+        time_window: Optional[str] = None,
+        filters: Optional[Dict[str, Any]] = None
+    ) -> Dict[str, Any]:
+        """
+        Aggregate transactions by account.
+        """
+        group_by = ["account_id"]
+        if account_id:
+            filters = filters or {}
+            filters["account_id"] = account_id
+            group_by = None
+        
+        return self.aggregate_transactions(
+            aggregate_function=aggregate_function,
+            field=field,
+            group_by=group_by,
+            filters=filters,
+            time_window=time_window
+        )
+    
+    def aggregate_by_time_period(
+        self,
+        aggregate_function: str,
+        field: str,
+        time_period: str = "hour",  # hour, day, week, month
+        group_by: Optional[List[str]] = None,
+        filters: Optional[Dict[str, Any]] = None,
+        date_from: Optional[datetime] = None,
+        date_to: Optional[datetime] = None
+    ) -> Dict[str, Any]:
+        """
+        Aggregate transactions by time periods with optional additional grouping.
+        """
+        # Add time period to group by
+        time_group_by = [f"time_{time_period}"]
+        if group_by:
+            time_group_by.extend(group_by)
+        
+        # Build query with time-based grouping
+        query = self._build_time_aggregation_query(
+            aggregate_function, field, time_period, group_by, filters, date_from, date_to
+        )
+        
+        # Execute query
+        start_time = datetime.utcnow()
+        results = query.all()
+        execution_time = (datetime.utcnow() - start_time).total_seconds() * 1000
+        
+        # Format time-based results
+        formatted_results = self._format_time_aggregation_results(
+            results, aggregate_function, field, time_period, group_by
+        )
+        
+        return {
+            "results": formatted_results,
+            "metadata": {
+                "aggregate_function": aggregate_function,
+                "field": field,
+                "time_period": time_period,
+                "group_by": group_by,
+                "filters": filters,
+                "execution_time_ms": execution_time,
+                "total_periods": len(formatted_results),
+                "computed_at": start_time.isoformat()
+            }
+        }
+    
+    def get_velocity_metrics(
+        self,
+        user_id: str,
+        time_window: str = "24h",
+        transaction_type: Optional[str] = None
+    ) -> Dict[str, Any]:
+        """
+        Get comprehensive velocity metrics for a user.
+        """
+        filters = {"user_id": user_id}
+        if transaction_type:
+            filters["transaction_type"] = transaction_type
+        
+        # Get multiple aggregations in parallel
+        metrics = {}
+        
+        # Transaction count
+        count_result = self.aggregate_transactions(
+            "count", "*", filters=filters, time_window=time_window
+        )
+        metrics["transaction_count"] = count_result["results"]
+        
+        # Amount sum
+        sum_result = self.aggregate_transactions(
+            "sum", "amount", filters=filters, time_window=time_window
+        )
+        metrics["total_amount"] = sum_result["results"]
+        
+        # Average amount
+        avg_result = self.aggregate_transactions(
+            "avg", "amount", filters=filters, time_window=time_window
+        )
+        metrics["average_amount"] = avg_result["results"]
+        
+        # Maximum amount
+        max_result = self.aggregate_transactions(
+            "max", "amount", filters=filters, time_window=time_window
+        )
+        metrics["max_amount"] = max_result["results"]
+        
+        # Unique devices
+        device_result = self.aggregate_transactions(
+            "count_distinct", "device_id", filters=filters, time_window=time_window
+        )
+        metrics["unique_devices"] = device_result["results"]
+        
+        # Unique channels
+        channel_result = self.aggregate_transactions(
+            "count_distinct", "channel", filters=filters, time_window=time_window
+        )
+        metrics["unique_channels"] = channel_result["results"]
+        
+        return {
+            "user_id": user_id,
+            "time_window": time_window,
+            "metrics": metrics,
+            "computed_at": datetime.utcnow().isoformat()
+        }
+    
+    def _build_aggregation_query(
+        self,
+        aggregate_function: str,
+        field: str,
+        group_by: Optional[List[str]],
+        filters: Optional[Dict[str, Any]],
+        time_window: Optional[str],
+        date_from: Optional[datetime],
+        date_to: Optional[datetime]
+    ):
+        """
+        Build SQLAlchemy query for aggregation.
+        """
+        # Start with base query
+        query = self.db.query()
+        
+        # Build SELECT clause
+        select_fields = []
+        
+        # Add group by fields
+        if group_by:
+            for group_field in group_by:
+                if hasattr(Transaction, group_field):
+                    select_fields.append(getattr(Transaction, group_field))
+        
+        # Add aggregation field
+        if aggregate_function == "count":
+            if field == "*":
+                agg_field = func.count()
+            else:
+                field_attr = getattr(Transaction, field, None)
+                if field_attr:
+                    agg_field = func.count(field_attr)
+                else:
+                    raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "sum":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.sum(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "avg":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.avg(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "max":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.max(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "min":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.min(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "count_distinct":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.count(func.distinct(field_attr))
+            else:
+                raise ValueError(f"Field {field} not found")
+        else:
+            raise ValueError(f"Unsupported aggregate function: {aggregate_function}")
+        
+        select_fields.append(agg_field.label('agg_value'))
+        
+        # Apply SELECT
+        query = query.add_columns(*select_fields).select_from(Transaction)
+        
+        # Apply filters
+        if filters:
+            for filter_field, filter_value in filters.items():
+                if hasattr(Transaction, filter_field):
+                    field_attr = getattr(Transaction, filter_field)
+                    if isinstance(filter_value, list):
+                        query = query.filter(field_attr.in_(filter_value))
+                    else:
+                        query = query.filter(field_attr == filter_value)
+        
+        # Apply time filters
+        if time_window:
+            time_delta = self._parse_time_window(time_window)
+            cutoff_time = datetime.utcnow() - time_delta
+            query = query.filter(Transaction.created_at >= cutoff_time)
+        
+        if date_from:
+            query = query.filter(Transaction.created_at >= date_from)
+        
+        if date_to:
+            query = query.filter(Transaction.created_at <= date_to)
+        
+        # Apply GROUP BY
+        if group_by:
+            for group_field in group_by:
+                if hasattr(Transaction, group_field):
+                    query = query.group_by(getattr(Transaction, group_field))
+        
+        # Add ORDER BY for consistent results
+        if group_by:
+            query = query.order_by(desc('agg_value'))
+        
+        return query
+    
+    def _build_time_aggregation_query(
+        self,
+        aggregate_function: str,
+        field: str,
+        time_period: str,
+        group_by: Optional[List[str]],
+        filters: Optional[Dict[str, Any]],
+        date_from: Optional[datetime],
+        date_to: Optional[datetime]
+    ):
+        """
+        Build query for time-based aggregation.
+        """
+        # Time period extraction
+        if time_period == "hour":
+            time_expr = func.strftime('%Y-%m-%d %H:00:00', Transaction.created_at)
+        elif time_period == "day":
+            time_expr = func.strftime('%Y-%m-%d', Transaction.created_at)
+        elif time_period == "week":
+            time_expr = func.strftime('%Y-W%W', Transaction.created_at)
+        elif time_period == "month":
+            time_expr = func.strftime('%Y-%m', Transaction.created_at)
+        else:
+            raise ValueError(f"Unsupported time period: {time_period}")
+        
+        # Start building query
+        select_fields = [time_expr.label('time_period')]
+        
+        # Add group by fields
+        if group_by:
+            for group_field in group_by:
+                if hasattr(Transaction, group_field):
+                    select_fields.append(getattr(Transaction, group_field))
+        
+        # Add aggregation field
+        if aggregate_function == "count":
+            if field == "*":
+                agg_field = func.count()
+            else:
+                field_attr = getattr(Transaction, field, None)
+                if field_attr:
+                    agg_field = func.count(field_attr)
+                else:
+                    raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "sum":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.sum(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        elif aggregate_function == "avg":
+            field_attr = getattr(Transaction, field, None)
+            if field_attr:
+                agg_field = func.avg(field_attr)
+            else:
+                raise ValueError(f"Field {field} not found")
+        else:
+            raise ValueError(f"Unsupported aggregate function: {aggregate_function}")
+        
+        select_fields.append(agg_field.label('agg_value'))
+        
+        # Build query
+        query = self.db.query(*select_fields)
+        
+        # Apply filters
+        if filters:
+            for filter_field, filter_value in filters.items():
+                if hasattr(Transaction, filter_field):
+                    field_attr = getattr(Transaction, filter_field)
+                    if isinstance(filter_value, list):
+                        query = query.filter(field_attr.in_(filter_value))
+                    else:
+                        query = query.filter(field_attr == filter_value)
+        
+        # Apply date filters
+        if date_from:
+            query = query.filter(Transaction.created_at >= date_from)
+        if date_to:
+            query = query.filter(Transaction.created_at <= date_to)
+        
+        # Group by time period and other fields
+        query = query.group_by(time_expr)
+        if group_by:
+            for group_field in group_by:
+                if hasattr(Transaction, group_field):
+                    query = query.group_by(getattr(Transaction, group_field))
+        
+        # Order by time period
+        query = query.order_by(time_expr)
+        
+        return query
+    
+    def _format_aggregation_results(
+        self,
+        results: List[Tuple],
+        aggregate_function: str,
+        field: str,
+        group_by: Optional[List[str]]
+    ) -> Union[Dict[str, Any], List[Dict[str, Any]]]:
+        """
+        Format aggregation results into a consistent structure.
+        """
+        if not group_by:
+            # Single result
+            if results:
+                return {"value": float(results[0].agg_value) if results[0].agg_value else 0}
+            else:
+                return {"value": 0}
+        
+        # Multiple results with grouping
+        formatted = []
+        for result in results:
+            result_dict = {}
+            
+            # Add group by fields
+            for i, group_field in enumerate(group_by):
+                result_dict[group_field] = getattr(result, group_field, None)
+            
+            # Add aggregated value
+            result_dict["value"] = float(result.agg_value) if result.agg_value else 0
+            
+            formatted.append(result_dict)
+        
+        return formatted
+    
+    def _format_time_aggregation_results(
+        self,
+        results: List[Tuple],
+        aggregate_function: str,
+        field: str,
+        time_period: str,
+        group_by: Optional[List[str]]
+    ) -> List[Dict[str, Any]]:
+        """
+        Format time-based aggregation results.
+        """
+        formatted = []
+        for result in results:
+            result_dict = {
+                "time_period": result.time_period,
+                "value": float(result.agg_value) if result.agg_value else 0
+            }
+            
+            # Add group by fields if present
+            if group_by:
+                for group_field in group_by:
+                    result_dict[group_field] = getattr(result, group_field, None)
+            
+            formatted.append(result_dict)
+        
+        return formatted
+    
+    def _generate_cache_key(
+        self,
+        aggregate_function: str,
+        field: str,
+        group_by: Optional[List[str]],
+        filters: Optional[Dict[str, Any]],
+        time_window: Optional[str],
+        date_from: Optional[datetime],
+        date_to: Optional[datetime]
+    ) -> str:
+        """
+        Generate cache key for aggregation.
+        """
+        key_parts = [
+            f"agg:{aggregate_function}",
+            f"field:{field}",
+            f"group_by:{':'.join(group_by) if group_by else 'none'}",
+            f"filters:{json.dumps(filters, sort_keys=True) if filters else 'none'}",
+            f"time_window:{time_window or 'none'}",
+            f"date_from:{date_from.isoformat() if date_from else 'none'}",
+            f"date_to:{date_to.isoformat() if date_to else 'none'}"
+        ]
+        return "|".join(key_parts)
+    
+    def _get_cached_result(self, cache_key: str) -> Optional[Dict[str, Any]]:
+        """
+        Get cached aggregation result.
+        """
+        cached = self.db.query(AggregateCache).filter(
+            AggregateCache.cache_key == cache_key,
+            AggregateCache.expires_at > datetime.utcnow()
+        ).first()
+        
+        if cached:
+            result = json.loads(cached.cache_value)
+            result["metadata"]["cache_hit"] = True
+            return result
+        
+        return None
+    
+    def _cache_result(
+        self,
+        cache_key: str,
+        result: Dict[str, Any],
+        time_window: Optional[str]
+    ):
+        """
+        Cache aggregation result.
+        """
+        # Determine cache duration
+        if time_window:
+            time_delta = self._parse_time_window(time_window)
+            # Cache for 10% of time window, min 1 minute, max 1 hour
+            cache_duration = max(min(time_delta * 0.1, timedelta(hours=1)), timedelta(minutes=1))
+        else:
+            cache_duration = timedelta(minutes=5)
+        
+        expires_at = datetime.utcnow() + cache_duration
+        
+        # Upsert cache entry
+        existing = self.db.query(AggregateCache).filter(
+            AggregateCache.cache_key == cache_key
+        ).first()
+        
+        if existing:
+            existing.cache_value = json.dumps(result)
+            existing.expires_at = expires_at
+        else:
+            cache_entry = AggregateCache(
+                cache_key=cache_key,
+                cache_value=json.dumps(result),
+                expires_at=expires_at
+            )
+            self.db.add(cache_entry)
+        
+        self.db.commit()
+    
+    def _parse_time_window(self, time_window: str) -> timedelta:
+        """
+        Parse time window string into timedelta.
+        """
+        if time_window.endswith('h'):
+            hours = int(time_window[:-1])
+            return timedelta(hours=hours)
+        elif time_window.endswith('d'):
+            days = int(time_window[:-1])
+            return timedelta(days=days)
+        elif time_window.endswith('m'):
+            minutes = int(time_window[:-1])
+            return timedelta(minutes=minutes)
+        else:
+            raise ValueError(f"Unsupported time window format: {time_window}")
+    
+    def cleanup_expired_cache(self):
+        """
+        Remove expired cache entries.
+        """
+        self.db.query(AggregateCache).filter(
+            AggregateCache.expires_at < datetime.utcnow()
+        ).delete()
+        self.db.commit()
\ No newline at end of file