Software Engineer - 1 at Avathon (formerly SparkCognition India Pvt. Ltd.) (2025-08 – Present)
Working on Visual AI Advisor (VAIA) and Central Asset Manager (CAM) platforms
- Replaced a monolithic C# backend with a modular FastAPI + Django ORM platform, reducing p99 API latency from ~400ms to <100ms via async handling and serving multi-tenant concurrent tenants
- Reduced report retrieval latency across multi-tenant tables by optimizing PostgreSQL queries (JOINs, window functions), eliminating inefficient scans and improving pagination performance
- Improved scalability and reliability of scheduled report generation by implementing a Celery + Redis async pipeline with a fan-out dispatcher, enabling retries and controlled concurrency across multiple workers
- Implemented observability & security: structured logging, request tracing, centralized exceptions, and Keycloak-based JWT/RBAC and Kubernetes health/readiness probes across 40+ APIs
- Developed and maintained a multi tenant asset management platform using Django REST Framework and PostgreSQL, serving as a central hub for managing organizations, users, assets, and hierarchies across tenants
- Architected a GraphQL API layer using Strawberry GraphQL with support for queries, mutations, and subscriptions, optimizing data retrieval patterns and enhancing real-time communication between services and frontend clients
- Implemented real time WebSocket handling with Django Channels, including connection lifecycle management and idle timeout cleanup to prevent resource leaks
- Strengthened platform security and access control by implementing Keycloak based JWT authentication with RBAC, custom middleware validation, impersonation logging, and webhook signature verification
- Set up CI/CD pipelines using Bitbucket Pipelines with (Dev, Sandbox, UAT, AWS) across cloud providers GCP and AWS, each with a 3-stage flow (Test → Docker Build → Helm Deploy), including containerization and pytest testing
