- Led the optimization of all reporting endpoints, successfully reducing latency by approximately 60% through advanced query tuning and improved indexing strategies, resulting in faster and more efficient data retrieval.
- Integrated Meta APIs to enable seamless connectivity and feature support for platforms under Meta, including WhatsApp, Instagram, and Facebook, enhancing product capabilities and user engagement.
- Migrated broadcast reporting infrastructure from Postgres to Metabase, utilizing AWS Athena as the data source to improve scalability, query performance, and analytical flexibility.
- Implemented various performance enhancements for Postgres queries, including the use of Common Table Expressions (CTE), multi-column indexing, and materialized views, significantly reducing response times across the platform.
- Addressed real-time chat notification delays by migrating from FCM to Vernemq WebSocket and developed an ACK (Acknowledgement) system for more reliable tracking and reporting of message deliveries.
- Designed and built a real-time chat counter system leveraging Kafka, Flink, and Clickhouse, providing accurate and high-throughput metrics for chat activities.
- Improved and consistently maintained the Chat Broadcast Service Level Objective (SLO) at 99.9%, ensuring high reliability and user satisfaction.
- Enhanced WhatsApp broadcast throughput by implementing dynamic throttling with Sidekiq Throttled, resolving persistent queuing issues, and successfully reducing Meta rate limit errors from over 1,054,871 to zero.
- Initiated the migration of Sidekiq and RabbitMQ workers from virtual machine (VM)-based infrastructure to Kubernetes, enabling better scalability, reliability, and operational efficiency.
- Resolved latency issues in the delivery of external webhook endpoints by re-architecting the webhook system into an event-driven architecture, introducing a scoring mechanism for endpoints based on latency, and segmenting worker queues according to these categories for optimal performance.
- Collaborated closely with the DevOps team to implement Kubernetes Horizontal Pod Autoscaler (HPA) using Datadog metrics and Keda, enabling automatic scaling of Sidekiq workers based on workload demands.
- Provided mentorship to several team members, supporting their individual initiatives and fostering a culture of continuous learning and professional growth.
- Developed a high-performance email broadcast service from scratch using Golang, Kafka, and Alicloud Direct Mail, utilizing Go routines for efficient multithreading.
Technologies: Ruby, Golang, Postgres, Elasticsearch, Redis, Kafka, RabbitMQ, Sidekiq, Vernemq, AWS, Alicloud, Kubernetes, Datadog, Grafana