Aiswarya Vijaya Kumar

Amazon Business • Restored query SLA compliance across 4 Redshift clusters by designing an idempotent Step Functions framework executing 600+ VACUUM and 3,200+ ANALYZE operations daily with retry logic, failure isolation, and throttle-aware execution. • Eliminated manual configuration and configuration drift by deploying the infrastructure via AWS CDK with IAM boundaries, observability, alerting, and per-cluster failure isolation for auditable, reversible operations. • Built QuickSight observability dashboards tracking VACUUM/ANALYZE completion rates, p95 query latency, and storage utilization, reducing on-call triage time from hours to minutes.

• Built a LangGraph text-to-SQL agent that generates and validates Snowflake queries grounded in dbt schema metadata, with an error-feedback loop, served via FastAPI to enable analysts to self-serve data requests, reducing turnaround from days to minutes. • Built a Snowpipe ingestion pipeline with S3 event triggers achieving sub-5-minute data freshness into Snowflake, backed by dbt transformations, instrumented with Grafana metrics and ElasticSearch logs to surface ingestion lag and test failures.

• Enabled daily e-commerce reconciliation and Tableau analytics for finance and ops stakeholders by orchestrating batch PySpark ETL pipelines via Airflow DAGs, processing high-volume workloads from ingestion to downstream consumption. • Cut Spark processing latency by 55% by profiling shuffle skew, partition sizing, execution plans and lowering compute costs. • Designed a star schema warehouse for e-commerce order analytics (order,sales facts, product/customer/time dimensions) supporting sub-second aggregation across 5TB+ of transactional data for business intelligence workloads. • Architected a microfrontend e-commerce platform using React and JavaScript, reducing initial page load by 65% and improving Core Web Vitals by 40% through lazy loading, efficient data fetching and tree shaking. • Built Python microservices deployed via Docker for order and inventory management, designing versioned RESTful APIs with OAuth, rate limiting, input sanitization, and structured error contracts, with near-real-time event propagation via Kafka. • Built GitLab CI/CD pipelines with blue-green deployments and expanded AAA-structured test coverage from 55% to 95%, adding pre-release translation drift detection that cut localization bugs by 30% before production.

Grade : 3.9/4.0 Course work : Operating Systems Advanced Computer Security and Privacy Big Data Systems Data Exploration, Cleaning and Integration Artificial Intelligence Trustworthy AI Computer Vision