Ashesh Srivastava
← home

QEG Automation — the machine that ran 77 storefronts

A managed-service business running US Walmart marketplace stores for client owners, fulfilled from Amazon, on a custom platform I had built and a ~35-person ops team hired in eight weeks.

FounderOps at scaleInternal platformHiring machine
Role
Founder — platform, team, operations, client economics.
Timeline
2022 — 2023 founder memory
Status
Sold founder memory.
Scale
77 client stores · 204,559 live listings · ~35-person ops team.
Tech
Custom PHP + MySQL platform · Amazon MWS + Walmart Marketplace APIs · competitive repricer · 14 cron jobs · order & tracking automation.
$593K
peak monthly GMV (Aug 2021) — GMV, not revenue
17,583
orders processed in that month
204,559
live listings under management

The business model was dropshipping. We created stores on Walmart.com and listed products priced 30–40% higher than the same items on Amazon.com. Whenever a confirmed order came in on Walmart, we placed that order on Amazon and had it shipped directly to the customer — the spread between the two prices was the margin. A typical dropshipping model; our innovation was the software we built to run it.

The platform

System map of the custom PHP automation platform: eight pipeline stages — catalog ingestion, ASIN to UPC resolution through two vendors in series, product enrichment via Amazon MWS, listing feed builder, competitive repricer, inventory sync, order pull, and tracking injection — flowing between Amazon's supply side and Walmart's marketplace APIs, backed by a 200k-row MySQL listing table and a 481,310-pair ASIN-UPC dataset.
The full pipeline, catalog to cash — eight stages between Amazon and Walmart, fourteen cron jobs, four firing every single minute.
The numbers, labeled honestly
Peak observable month: August 2021 — $593,653 in Walmart GMV on 17,583 orders, with a $48,637 peak day. Behind that volume was a team of 35 managing every one of those orders day to day, with supervisors tiered above them overseeing the operation.
Dark-themed QEG Automation operations console for August 2021: $593,653 month-to-date GMV, 17,583 orders, 204,559 live listings, 75 of 77 stores active, 98.4% tracking injected, a daily GMV chart with the $48,637 August 8 peak, top stores by 30-day GMV, the cron-job pipeline table, a 77-cell store fleet grid, and the latest orders with per-order spread.
The operations console, reconstructed — August 2021 at peak, every number from the production database.

The learnings

  1. Complex systems are built in the long tail. The core pipeline worked early; the long-tail edge cases took the longest by far — that last stretch is most of what "production-ready" actually means.
  2. A business that relies 100% on one platform is susceptible to collapse. Every listing, order, and dollar routed through Walmart's marketplace; when the channel turned, the whole machine stopped with it.
  3. Build redundancy for every API you depend on. External calls fail — the ASIN→UPC step ran two vendors in series because neither alone had coverage, and every critical dependency needs that kind of fallback.
Next: QuicSnap → Shutter Labs →