Backend platform for upGrad's learning, exams and ERP systems

Context

upGrad is an EdTech company. I am the technical owner for several of its mission-critical backend systems, all built in Spring Boot (Java):

• Referral Service: learner referrals.
• Exam Configurator: setting up and configuring exams.
• Gradr: the exam proctoring platform.
• A large-scale ERP: internal operations.

As owner I make the architecture, service-boundary, scalability and reliability decisions for these systems.

Design and data modeling

I write the high-level designs and database schemas for new backend services. That covers relational stores (MySQL, PostgreSQL) and Elasticsearch, with a focus on data modeling and performance at scale.

Performance

I added Redis-based caching to hot read paths. It cut API latency, raised throughput and took a large share of the load off the primary databases.

Event-driven communication

I brought in Kafka for asynchronous, event-driven communication between microservices. Services stay loosely coupled, and a failure in one service no longer spreads to the services that depend on it.

Real-time features in the LMS

• WebSockets for live learner–faculty interaction.
• Server-Sent Events (SSE) for real-time notifications inside the learning platform.

Payments

I designed idempotent APIs for the payment workflows. A retried or duplicated request cannot charge a learner twice or leave a payment in an inconsistent state, which matters in a distributed system where retries are normal.

Production health

I track down production performance issues with Grafana, Instana and AWS Athena, and use what I find to choose which optimizations to ship.