Yash Mathur

Experienced Backend Engineer specializing in building scalable systems for SaaS, Ed-Tech, and Fin-Tech industries. Skilled in Node.js with a focus on distributed systems, microservices architecture, and cloud deployments. Led backend projects at Kurated.ai, integrated observability tools like Datadog, Prometheus, and PagerDuty, and optimized system performance at Regenesys and Creative Galileo. Created Go-Signal, an in-app notification system using Golang, Kafka, Cassandra, and Redis. Proficient in designing resilient APIs, managing messaging systems, and driving DevOps processes.

GoSignal – Scalable In-App Notification System

• Built as a plug-and-play notification system for startups and individuals, similar to LinkedIn’s in-app notification tab.
• Developed using Golang to ensure high concurrency and performance.
• Designed for easy integration into existing platforms with minimal setup.
• Uses Apache Kafka for scalable message queuing and retry logic for failed notifications.
• Integrated Cassandra to maintain point-to-point tracking of delivered messages.
• Stores user socket/session info in Redis to ensure precise and real-time message delivery.
• Supports multi-instance deployments, with inter-instance communication via gRPC or REST.
• Logs and monitors activity using Datadog, ensuring observability and easier debugging.
• Built with a modular strategy, allowing components like queue, cache, or DB to be swapped without major changes.
• MySQL (via GORM) used as the core DB with flexibility to switch to other relational databases.

Notificania – One system to deliver every message

• Built Notificania, a messaging platform capable of sending various message types:
  Emails .
  Firebase Push Notifications.
  Apple Push Notifications (upcoming).
  SMS .
  WhatsApp Messages (in development).
• Developed the backend using Node.js.
• Integrated Apache Kafka as a message broker to handle high message loads and support horizontal scalability.
• Designed the system to support multiple instances for distributed load handling.
• Built a highly configurable and fully extensible architecture—e.g., implemented Twilio for SMS but designed it to easily support alternative providers in the future.
• Used Apache Cassandra for point-to-point message tracking.
• Utilized MySQL as the core database, with support for switching to other relational databases.
• Achieved 506.2 requests/sec performance during a JMeter load test with 12,000 mail requests.
• Integrated Redis for caching to improve message delivery efficiency.
• App source code: https://github.com/themockingjester/notificania

• Human Scream Detection Through Three-Stage Supervised Learning and Deep Learning (08/2020 - 08/2020) - this research paper tells us how noises from the environment can be analysed using machine learning and deep learning models for decreasing crime rate. here is the link.
• Served as a Founding Engineer - I have served as a founding engineer in startups (like kurated.ai)