rahul verma

Architected and independently delivered a distributed data-archival service after POCs of self-managed, MongoDB Atlas Data Federation, and AWS Glue, streaming multi-TB cold data to S3. Deployed across all clusters, archiving ~60% historical data and enabling M60→M40 downgrade, cutting DB costs ~35% ($4–5K/month). Optimized 15+ collections, reducing p95 latency 40% and CPU 25%. Improved payment reconciliation 30%, saving 10+ hrs/week. Used Copilot/Claude Code to cut dev time 20%.

Developed automated Darwinbox leave management for 3,000+ trainers and streamlined onboarding, cutting operational effort 60% and ramp-up time from 5 to 2 days. Built performance tracking and feedback loops, increasing trainer engagement 20%. Led improvements across roster, cult-api, and fitness-management, scaling to 100K+ daily requests across 150+ centers. Optimized APIs with structured logging and async flows, reducing dashboard load time 30%.

Built a Drools-based rule engine and rules-based approval engine, and integrated Javers for audit trails, reducing review time by 40%. Introduced TestContainers for org-wide integration testing and upgraded KafDrop with filters and caching, reducing debugging and API response time by up to 80%.

Built Pack Box Master, saving ~$3,600/year and 1 hour/day in packing time, and resolved concurrency issues to stabilize backend flows. Load-tested services to support 10k+ daily operations and automated test pipelines, saving 40+ QA hours per cycle while mentoring a 3-member team.