RHRipan Halder Résumé ↓

Professional case study

TVG Racing microservices.

High-traffic backend systems spanning wagering, wallets, payments, real-time racing data, gateways and operational tooling.

Java 17Spring Boot 3KafkaAWSRedisPostgreSQLGraphQL
27Services in modernization scope
4Delivery environments
25%Approximate build-time reduction
0Critical outages in supported marquee windows

Context

The platform uses focused domain services rather than one large application. Client traffic passes through controlled gateways, while authentication, account, wallet, wagering, payment, content and promotions capabilities remain independently deployable and scalable.

Ripan’s contribution

Ripan worked on Spring Boot 3 services during the 2024–2025 FanDuel engagement, contributing to service reliability, delivery modernization, cloud migration, operational visibility and collaboration with product and frontend teams. The broader modernization scope covered 27 services across four environments.

Architecture patterns

Web / MobileAPI GatewayDomain servicesKafka / RabbitMQOwned data

Synchronous REST or GraphQL calls handle operations that require an immediate answer. Asynchronous messaging handles workflows that benefit from buffering, retries, failure isolation or delayed completion. This split helps prevent slow downstream systems from blocking the client.

Read and write separation

Transactional services focus on authoritative writes, while read-oriented services and Redis-backed views absorb frequent status and history queries. The design reduces contention and lets read traffic scale independently from live transaction processing.

Failure-aware integrations

External payment, tote, geolocation and content providers are isolated behind gateways or dedicated services. Timeouts, retries, asynchronous status flows and clear ownership reduce the blast radius of third-party degradation.

Observability

Correlation IDs, centralized logs, metrics and queue-lag monitoring make it possible to follow a request across service boundaries. This is especially important when a user-facing operation becomes asynchronous and completes through multiple consumers.

Measured impact

  • Approximately 25% faster builds through build and dependency modernization.
  • Four environments kept aligned through Gradle, JFrog, Vault and cloud migration work.
  • A check-withdrawal path had a historical reference of roughly 2,400 requests per hour at a 2021 Kentucky Derby peak and was validated to 6 requests per second in QA using two Kafka partitions. This is system context, not a claim that Ripan worked there in 2021.
  • No critical outages during the Kentucky Derby and Breeders’ Cup windows Ripan supported.

Engineering lessons

The most important lesson is that scale is not only about adding instances. Stable service boundaries, explicit data ownership, asynchronous workflows, observability and operational playbooks matter as much as raw compute.