A'sTechware Logo, AI & Platform Engineering
A'sTechware Logo, AI & Platform Engineering
© A'sTechware / All rights reserved.

A'sTechware Logo, AI & Platform Engineering
Menu
Custom Software & AI for Operations
Book a 30-Min Strategy Call
Share

High-Performance Search & Inventory Recovery

Restaurant Resource Database Rescue

An emergency performance audit and search rebuild for a large supplier platform—designed to handle complex part numbers, special characters, and real-time inventory updates.

The Challenge

A B2B industrial restaurant supplier’s search engine was effectively broken. With a large catalog, searching for a specific part took too long, driving massive cart abandonment. A previous vendor failed to implement a search architecture that could handle complex part numbers with special characters.

The instability wasn’t isolated to search: background processing leaked memory, the admin experience degraded under load, and the platform crashed predictably. The rescue required a foundation-first approach: stabilize runtime behavior, fix query patterns, then rebuild search as a dedicated system.

Quick Stats

  • Catalog: Large SKU catalog
  • Search: Meilisearch (typo-tolerant)
  • Ops: Sidekiq stabilization
  • Impact: Faster search and improved stability through performance and queue hardening

The Solution

We performed an emergency performance audit and replaced the legacy SQL LIKE search with a dedicated in-memory search engine. We also refactored the ingestion pipeline to support real-time inventory updates from multiple supplier feeds.

The outcome was a stable, fast platform: sub-second part lookup even with typos and colloquial terms, predictable background processing, and an admin workflow that no longer degrades during daily operational peaks.

Technical Approach

  • In-memory indexing: Migrated search to Meilisearch for ultra-low latency and typo tolerance.
  • Memory management: Fixed leaks in background processing that were crashing servers on a predictable cadence.

Technical Details

Architecture

Ruby on Rails → Sidekiq → Meilisearch

Integrations

Inventory management API + custom vendor feeds (CSV/XML).

Security

Secure API gateway; daily encrypted backups (30-day retention); RBAC admin dashboard.

AI Features

Semantic matching maps colloquial searches to the correct technical part and suggests compatible items.

Results & Impact

  • Faster search results through dedicated indexing and typo-tolerant lookup.
  • Improved uptime through platform stabilization and safer background processing.
  • Faster admin workflows for order processing and internal operations.