Case Study · Marketplace

Techjockey: US & UAE Marketplace Launch

End-to-end architecture and delivery of two international SaaS marketplaces — ₹2 Cr revenue in the first launch cycle.

Role: SDE-III & Tech Lead · Team: 6-engineer pod · Timeline: 2023 launch cycle · Stack: Node.js · TypeScript · AWS · MySQL · Redis · Docker

High-level architecture: India / US / UAE traffic routed through region-aware API gateways into a shared service layer with locale-aware pricing and region-specific payment integrations.

The Problem

Techjockey, India's largest B2B SaaS marketplace, wanted to expand into the US and UAE — two markets with very different tax, currency, payment, and compliance constraints. The existing India-only stack was tightly coupled to local payment rails and assumed INR pricing throughout the data model.

The asks weren't subtle. US listings needed dollar pricing, US sales-tax handling, and Stripe/PayPal-class payment flows. UAE listings needed AED pricing, VAT inclusivity, and regional gateways with their own retry semantics. Customers in either region couldn't see Indian-only SKUs, and partners couldn't list a product to one region without spilling into another.

And the launch had a fixed window. We needed a path that didn't fork the codebase but also didn't ship a half-baked abstraction. The answer had to work for these two regions and not collapse when a third arrived next year.

The Solution

I led system design end-to-end: a multi-region API architecture with locale-aware pricing, region-specific payment gateway integrations, AWS-native deployment, and a CI/CD pipeline that let us promote builds across regions safely. I owned the REST API layer, set up the AWS infrastructure (Lambda, EC2, S3, load balancing), and partnered with frontend on the storefront work.

  1. Region as a first-class dimension. Every product, price, tax rule, and payment method became scoped by region in the data model. No more implicit INR or implicit India.
  2. Locale-aware pricing service. Pricing moved out of the product row into a dedicated service that knows about currency, rounding rules, and tax inclusivity per region.
  3. Pluggable payment integrations. A single payment-orchestrator interface, with region-specific adapters behind it. Adding a third region becomes a new adapter, not a new fork.
  4. AWS-native deploy. Lambda for the lighter region-specific handlers, EC2 for the core APIs, S3 + CloudFront for storefront assets, ALB for routing. CI/CD promotes a single artifact across region targets with per-region config.

Tech Stack

  • Node.js
  • TypeScript
  • MySQL
  • Redis
  • AWS · Lambda
  • AWS · EC2
  • AWS · S3
  • Docker
  • CI/CD

My Specific Contribution

  • System design. HLD and LLD for the multi-region architecture; reviewed every component design with the pod.
  • REST API design and implementation. Region-scoped endpoints, idempotency for payment paths, contract docs.
  • AWS infrastructure setup. Lambda functions, EC2 fleet, S3 buckets, ALB, IAM, environment isolation.
  • CI/CD pipeline. Single-artifact promotion across region targets, per-region config, automated rollback hooks.
  • Code review and standards. Reviewed every PR going to the release branch; set the bar for tests, observability, and rollback safety across the pod.

Impact

₹2 Cr

Revenue against a ₹3 Cr target in the first launch cycle.

60% ↑

Deployment efficiency lift after AWS migration; manual release errors eliminated.

0

Rollback incidents in launch quarter, across both regions.

Visuals

Deploy time before vs after the AWS migration — manual on-prem releases gave way to automated CI/CD with environment promotion.
Region-specific storefronts — locale-aware pricing, currency, and payment rails served from a shared service layer.

Lessons

The biggest lesson wasn't technical — it was that "internationalisation" means a hundred small decisions, not one big switch. Currency rounding, tax inclusivity, payment retry semantics, even the order of fields in a checkout form — each of these had a region-specific answer we hadn't thought about until a customer hit it. The right architecture isn't the one that abstracts those decisions away; it's the one that makes them cheap to change.

Like this work? Let's talk.

I'm open to Senior / Staff / Tech Lead roles at product companies — remote or Delhi-NCR hybrid.