PRAGYA AGARWAL

I built Python data ingestion pipelines for a clinical research platform processing 10K+ records per batch from AWS S3, with normalization and failure handling for inconsistent datasets. I engineered a field-level data reconciliation system across distributed study modules that automatically detected inconsistencies and reduced manual validation effort, and automated data quality workflows across ingestion and downstream processing to improve reliability of research datasets used by scientists.

I architected Manticore Console, a multi-tenant observability platform for GPU Cloud Service, serving enterprise customers via a scheduler API managing isolated Kubernetes clusters. I designed Django REST APIs for cluster lifecycle and workload visibility, and built React/D3 dashboards visualizing 10K+ infrastructure data points with reduced UI latency. I integrated UI, scheduler API, and control plane services with authentication, billing, orchestration, Tavern tests, CI/CD, and reusable components.

At the University of Texas at Arlington, I analyzed 50+ Java libraries and frameworks to identify security vulnerabilities across third-party dependency chains in enterprise applications, uncovering critical exposure patterns. I developed mitigation strategies and hardening guidelines for 20+ vulnerabilities that were adopted as standard practices by the research group to improve software security.

At Copart, I built Kafka-based event workflows that increased auction platform throughput by 85% for high-traffic bidding flows. I optimized backend data access by migrating Cassandra to MariaDB through Python ETL pipelines, reducing latency by 20%, and delivered Java/Spring Boot + Angular/TypeScript features for real-time bidding while improving code quality with JUnit and Maven pipelines.

At DXC Technology, I developed Java/Spring Boot microservices for insurance systems maintaining 99.8% availability with REST APIs, Spring Cloud, OAuth2, and JWT. I improved performance by 30% using async processing and multithreading, and containerized 50+ services with Docker/Kubernetes on AWS, reducing deployment cycles by 75%. I refactored React workflows, reducing manual errors by 60%, owned SLA-bound incident resolution, and mentored 20 engineers, reducing onboarding time by 65%.