Kumud Sharma

About

Backend engineer who has delivered large, measurable performance wins (10x throughput, 67% latency reduction) by combining Flask microservices, Redis caching, and AWS autoscaling/observability. Has hands-on depth in SQLAlchemy/Postgres optimization and production scaling pitfalls (cache consistency, connection exhaustion), plus experience deploying real-time ML inference (XGBoost) on AWS Lambda and building secure multi-tenant Kubernetes isolation.

Experience

Education

Key Strengths

• Built Flask microservice replacing legacy eligibility module; achieved 10x throughput and 67% latency reduction
• Designed Redis caching with write-through strategy and TTL tuning to resolve cache consistency issues
• Tuned AWS Auto Scaling (CloudWatch thresholds, lifecycle hooks, cooldowns) to stabilize scaling and reduce infra cost by 30%
• Solved MySQL connection exhaustion via robust connection pooling at scale
• PostgreSQL/SQLAlchemy performance optimization: composite indexing using EXPLAIN ANALYZE (~40% faster queries)
• Eliminated N+1 ORM issues with eager loading strategies; mitigated cartesian products via selectinload and reduced over-fetching with defer()
• Deployed real-time ML scoring service (XGBoost) behind Flask API on AWS Lambda; handled package limits with Lambda Layers
• Implemented cache versioning to safely roll out updated ML models without stale inference results
• Designed multi-tenant isolation on Kubernetes using namespaces, resource quotas, network policies, and RBAC
• Built reliable async data sync via Celery + Redis with exponential backoff retries and CloudWatch monitoring