New: The State of AI Assurance 2026 is out — download it free.

Retail & E-commerce / Performance Engineering

5× the scale, zero extra infrastructure — performance-engineering India’s largest e-commerce platform

India’s largest e-commerce platform · Retail & E-commerce

scalability (1,000 → 5,000 users, no added infra cost)
order processing (30 → 60/sec)
30%
backend performance gain
10%
more impressions per page
The context

India’s largest e-commerce platform serves millions across categories, peaking at its Big Billion Days sale. A seamless experience under extreme traffic is directly tied to revenue.

The challenge
  • Performance flatlined at 1,000–1,500 concurrent users due to autoscaling limits.
  • The warehouse order queue processed only 30–40 orders/second.
  • CPU bottlenecks cut front-end impressions by 10%.
What we did

Two-phase, scenario-based load testing on the client’s in-house Locust PaaS/IaC — exposing and removing the bottlenecks.

  • Simulated traffic surges; auto-scaled downstream systems behind load balancers.
  • Analysed GPU overdraw and trimmed network payloads.
  • Built a reusable API framework and optimised the order-confirmation API.
Draft — pending client approval
Going into Big Billion Days, we needed to know the platform would hold. They took us to 5× scale without adding a rupee of infrastructure.
Performance lead
Stack & tooling
LocustLoad balancers

Want outcomes like this?

Tell us where quality is slowing you down — we'll scope it in one call, outcomes defined upfront. Or run your own AI Exposure Snapshot in minutes.