Lakshay

Own a real-time order-tracking and data-collection platform end-to-end on Node.js, NATS, MongoDB, and React; take features from requirements to production without a dedicated PM, processing 10K+ events per day. Built fault-tolerant ingestion pipelines with retries, proxy rotation, and rate limiting; achieved 99.5%+ pipeline uptime even under partial upstream failures. Optimized MongoDB indexing and query patterns, reducing data-retrieval latency by 40%+ on the most-hit dashboards (p95 dropped from ~800ms to ~480ms). Designed and shipped React + TypeScript dashboards with reusable components and real-time WebSocket updates, used daily by internal operations team for live tracking and analytics.

Built and scaled mission-critical loan-processing and trade-simulation services in Java (Spring Boot), Python (FastAPI), and Go, serving an internal user base of analysts and external loan applicants. Designed a Go-based fraud-detection engine on Kafka streams, goroutines, and Redis, processing ~1M financial events per day at sub-200ms p99 latency. Developed a high-frequency trading simulator in Python and Go that consumes real-time L2 order-book data over WebSocket and applies Almgren-Chriss execution, slippage, and fee-prediction models for backtesting. Reduced API latency by 35%+ on the loan-processing path through async programming (asyncio, FastAPI), optimized SQL with proper indexing, Redis caching, and database connection pooling. Designed idempotent payment and loan APIs (idempotency keys, deduplication, exponential-backoff retries) to prevent double-charges and duplicate writes under network failures. Instrumented full-stack observability with Prometheus metrics, Grafana dashboards, and ELK-based centralized logging; reduced on-call mean-time-to-diagnose (MTTD) by ~50%.

Built and maintained logistics microservices in Java (Spring Boot) across order, inventory, and tracking domains; migrated time-sensitive modules to Go, reducing CPU usage by 45% and improving throughput by ~2x. Built a real-time notification system using the WhatsApp Business API, Kafka events, and Redis-based schedulers with retry and throttling; delivered ~100K notifications per day with 98%+ delivery success. Set up observability tooling (Prometheus + ELK Stack) across 8+ services, cutting debugging time by ~40% during production incidents. Improved order-tracking reliability and reduced return-to-origin (RTO) rate by 33% through predictive logic and address-validation checks, directly saving operational cost.

Built a dynamic brochure-generation system serving 25,000+ schools on a Django backend with optimized MySQL queries and template-based PDF rendering.