Skip to content
Devesh Singh
← All work

AdTech

Modcart

Full Stack DeveloperJun 2022 – May 2023modcart.io

Starting point

Joined a product that was already running

The platform was live with real advertisers on it. I joined to build new features, then took on the performance work.

What I owned

  • Designed
  • Built
  • Tested
  • Shipped
  • Operated

The moving parts

What talks to what — clients, services, data, and the boundaries between them.

IdentitySign-in providers
Advertiser web appClients
One-page storesEmbedded
Payments · MetaThird-party APIs
BackendServices
Shoppable adsEmbedded
DatabaseData
Event streamServices
CacheData

Push to prod

The checks a change has to pass before it ships, and what watches it after.

CI/CD
Test
Build
DeployFE · BE · Multi-env
Monitor

Where it lives

The cloud it runs on — ingress, compute, data and messaging — and how that gets provisioned.

AWS infrastructure
Infra as codeDelivery
Edge & DNSInfrastructure
Access controlInfrastructure
ComputeInfrastructure
Object storageData
KafkaOn EC2
CloudWatchOperations

Node detail

Pick a box

Tap any box in the diagrams to see the tools behind it and what each one did on this project. Tap again to let go.

The product

Advertising platform with embeddable single-page stores and real-time campaign analytics.

What I built

Joined as a full-stack developer (frontend and backend) on a team of 3 developers, 1 QA and 1 project lead, and took on DevOps after 6 months. Carried the platform's major enhancement work: designed the architecture and schemas for the interactive single-page stores, built access control, Razorpay payments and Kafka-backed analytics, took initial load from about 4 seconds to under 1, added a WebSocket inbox with Meta integrations, and defined every environment as code with CloudFormation, automating the deployment process — including the blue-green cutover that took deployment downtime from about one minute to under five seconds.

Everything I did here

  • Joined as a full-stack developer on a team of 3 developers, 1 QA and 1 project lead, working across frontend and backend, and took on DevOps responsibilities after 6 months.
  • Delivered significant enhancements to the advertising platform to boost user engagement and extend functionality.
  • Designed the architecture and database schemas for user-interactive single-page stores.
  • Built responsive single-page ad units that adapt across mobile, tablet and desktop, reaching 98% screen compatibility and letting advertisers reuse one creative across placements.
  • Cut initial load from about 4 seconds to under 1 second, so an embedded store renders before a visitor gives up on the host page.
  • Integrated robust user access management and payment systems using Node.js, Redis and MongoDB.
  • Integrated Razorpay for both platform payments and service payments across the advertiser flows.
  • Added Google and Meta social login, so advertisers onboard without creating another password.
  • Built role-based access control, payment integration, and analytics on ad interactions and campaign performance.
  • Developed dashboard analytics and ad interaction reports using Kafka, Redis and Node.js for real-time data processing and detailed decision-making insight.
  • Introduced Kafka event processing to decouple analytics ingestion from transactional workloads, with Redis caching for fast aggregation and reporting.
  • Implemented WebSocket communication and third-party tracking systems for real-time interaction with third-party social apps.
  • Built a WebSocket real-time chat with Meta integration — WhatsApp, Facebook and Instagram — so advertisers could reach customers across channels from a single inbox.
  • Defined dev, staging and production environments as code with CloudFormation, covering the admin portal, user portal, analytics service and ad delivery service.
  • Automated the deployment process and added blue-green deployment, cutting deployment downtime from roughly 1 minute to under 5 seconds.
  • Implemented CI/CD pipelines for consistent releases across the three environments.
  • Supported the entire infrastructure on AWS for scalable, secure deployment.
Download PDF