Integrated MCA student at Amrita Vishwa Vidyapeetham. I don't build demos - I build systems, then find out how they fail under real load, real concurrency, and real network partitions, and fix what I find before anyone else has to. Most recently: a 6-service microservices platform with an AI agent and a native MCP server, benchmarked to ~168,000 requests with zero failures.
Currently deep in resilience engineering, event-driven architecture, and applied AI infra (tool-calling agents, RAG, MCP).
shortlyai - Production-grade distributed URL-shortening platform built as a 6-service Spring Boot microservices system. Kafka SAGA choreography, transactional outbox, dual-layer circuit breakers, full Prometheus/Grafana/Loki/Tempo observability, an AI agent, and a native MCP server.
1,325 req/s · ~7ms average · 16.84ms p95 · 167,966 requests · 0 failures · 200 concurrent users
spring-ai-rag-service · spring-ai-chat-service · spring-ai-task-manager · spring-ai-mcp-server - four separate Spring AI services, each isolating one architectural pattern (streaming chat, scoped tool-calling, RAG, MCP) instead of one monolithic AI demo. All four enforce a human-in-the-loop safety boundary on destructive operations - the AI proposes, it never executes unilaterally.
Food-Waste-Management - Load-tested to 8,000 requests / 100 concurrent users / 0.00% error rate, designed explicitly around cache-outage and async-failure edge cases, not just the happy path.
