Few-Shot Vector Database Query Optimization

Optimize vector database queries based on these examples.

Example 1 - Filter Pushdown:
Original: Query all, then filter by metadata
Optimized: Pre-filter with metadata before vector search
Improvement: 3x faster, reduced scanned vectors by 80%

Example 2 - Batch Queries:
Original: 100 sequential single queries
Optimized: Batch into 10 requests of 10 queries each
Improvement: 5x throughput increase, 60% latency reduction

Example 3 - Index Partitioning:
Original: Single large index with 10M vectors
Optimized: Partition by tenant_id into smaller indices
Improvement: p99 latency reduced from 200ms to 50ms

Now optimize:
Current Query Pattern: {{query_pattern}}
Performance Issue: {{performance_problem}}
Database: {{vector_db}}