Search Performance

// Good - only fetch what you'll display
const results = await toolIndex.search(query, { topK: 5 });

// Wasteful - fetching more than needed
const results = await toolIndex.search(query, { topK: 1000 });

// Filter first, then rank - more efficient mental model
const results = await toolIndex.search(query, {
  filter: (m) => m.owner === 'billing',
  topK: 10,
});

const db = new VectoriaDB({
  useHNSW: true,
  hnsw: { efSearch: 50 },
});

const queryCache = new Map<string, SearchResult[]>();

async function cachedSearch(query: string) {
  const cacheKey = query.toLowerCase().trim();

  if (queryCache.has(cacheKey)) {
    return queryCache.get(cacheKey)!;
  }

  const results = await db.search(query);
  queryCache.set(cacheKey, results);

  // Clear cache after 5 minutes
  setTimeout(() => queryCache.delete(cacheKey), 5 * 60 * 1000);

  return results;
}

async function warmupSearch() {
  await db.initialize();

  // Warmup embedding pipeline
  await db.search('warmup query', { topK: 1 });

  console.log('Search ready');
}

// Instead of running multiple searches
const allResults = await db.search(query, { topK: 100 });

// Then filter in memory
const billingResults = allResults.filter(r => r.metadata.owner === 'billing');
const userResults = allResults.filter(r => r.metadata.owner === 'users');

async function monitoredSearch(query: string, options?: SearchOptions) {
  const start = Date.now();

  const results = await db.search(query, options);

  const latency = Date.now() - start;

  console.log({
    query: query.substring(0, 50),
    latencyMs: latency,
    resultCount: results.length,
    topScore: results[0]?.score,
  });

  return results;
}