Led enterprise-wide migration of 12+ mission-critical aviation apps from Spring Boot 2.7 → 3.x, cut deploy times 35% and eliminated 100+ deprecated dependencies across crew-scheduling and trip-trade platforms serving 15,000+ daily crew members.

Executed Hibernate 5 → 6 upgrade across 150+ entity mappings, refactoring N+1 query patterns for 40% faster crew legality checks and 28% improvement in sequence lookup performance.

Designed Multi-Sequence Trade microservices enabling 10,000+ monthly trip-sequence swaps with dynamic FAA duty-rest validation, cutting manual processing time 60%.

Led backend rearchitecture of a real-time departure-tracking console for Ops Control and dispatch, moving from polling-based updates to Kafka event streams — pushing gate and time changes to 2,000+ concurrent operators within 500ms and eliminating dashboard-refresh lag during peak IROPS.

Built the alerting engine for the same platform, correlating flight-plan updates, crew-legality events, and irregular-operations signals to auto-surface at-risk departures 30+ minutes earlier, cutting manual dispatcher scanning by 40% and reducing missed-connection cascades.

Implemented zero-trust security with OAuth2/OIDC and JWT across 30+ internal APIs spanning flight operations and crew management domains, achieving SOC 2 Type II compliance.

Delivered LLM-powered incident RCA copilot for the crew-ops on-call team, ingesting distributed traces, PagerDuty context, and internal runbooks to draft root-cause hypotheses with cited evidence — cut MTTR on P2 incidents from ~45 min to under 15 min.

Shipped an AI code-review agent (Claude + GPT hybrid) into the crew-scheduling CI pipeline, catching 60+ security or correctness issues per month that human reviewers were missing, with inline suggested diffs and repo-aware context.

Prototyped a RAG-based regulatory assistant over 14 CFR Part 117 and union CBAs that answers dispatcher and crew-planner questions with cited source paragraphs, now in internal beta with 50+ users across operations.

Delivered 6 microservices in the Claims domain (Intake, Validation, Notification) as part of the “One Digital Aflac” modernization, collectively handling 25K+ daily claim submissions.

Built Spring Boot 2.6/2.7 + Hibernate REST APIs for automated benefit calculation and policy rule validation (Accident, Cancer, Hospital Indemnity), cutting manual claim review workload by 40%.

Wired Kafka producers and consumers to sync claim status changes between backend services and customer portals within 2–3 seconds, reducing “where's my claim?” call-center inquiries by 25%.

Enhanced Spring Batch nightly claims jobs with chunk-oriented and parallel step execution, improving throughput 30% while handling 105K+ daily transactions.

Applied Spring Security + OAuth2 with field-level encryption for SSN and medical data across all endpoints, meeting HIPAA compliance for PHI handling.

Containerized services with Docker, deployed to AWS ECS/Fargate, contributed to Jenkins CI/CD supporting Aflac's on-prem-to-AWS migration.

Shipped backend service enhancements for the Amazon Business Procurement-to-Pay (P2P) workflow, improving enterprise-order approval-transition logic.

Optimized 5+ high-traffic REST APIs, cutting p99 latency 15% on the Guided Buying compliance engine via improved DynamoDB query patterns.

Built event-driven components on Amazon EventBridge + SQS to automate real-time B2B account status updates across distributed regional services.

Authored AWS CDK constructs automating provisioning of SQS queues, Lambda functions, and DynamoDB tables — enforced dev/prod parity.

Extended the team's AWS CDK library with a reusable L3 construct that wraps CodeDeploy applications, CloudWatch alarms, and SNS notification topics into a single canary-ready stack, cutting new-service deployment scaffolding from ~200 lines of CDK boilerplate to ~20.

Wrote CDK unit tests with the aws-cdk-lib/assertions module to verify canary stack outputs (weighted target groups, alarm thresholds, rollback triggers), catching a misconfigured error-rate threshold in code review before it reached the deployment pipeline.

Implemented canary deployment strategy with AWS CodeDeploy, shifting production traffic 10% → 50% → 100% for the P2P service and preventing two regressions from reaching full customer traffic during my internship.

Configured CloudWatch Synthetics canaries probing critical B2B checkout endpoints every 5 minutes, wiring alarms to automated CodeDeploy rollback triggers on elevated p99 latency or 5xx error rates.

Contributed AWS Glue scripts to an S3-based Data Lake pipeline transforming raw B2B transactional logs into queryable spend-analytics datasets.

Achieved 95% code coverage on payment-integration modules with rigorous JUnit + Mockito unit and integration tests.

Designed Spring Boot microservices for AT&T, P&G, and PepsiCo engagements, processing 500K+ daily transactions with REST APIs documented via Swagger/OpenAPI.

Built AT&T customer analytics platform (Spring Boot 2.3 + React + Oracle) processing millions of interaction records for churn indicators and executive dashboards.

Architected Cassandra-based time-series storage for real-time consumer sentiment analytics, managing 5–10TB distributed datasets with sub-second query performance.

Delivered event-driven architecture with JMS + ActiveMQ for decoupled data ingestion, processing 100K+ daily messages with guaranteed delivery.

Built high-throughput ETL pipelines with Spring Batch + Hibernate handling 100K–5M+ records daily, using ExecutorService for 3× performance improvement.

Automated JasperReports generation for 30–50 weekly client reports with dynamic data binding and parameterized templates.

Built Cox Communications' customer self-service portal (Spring MVC + Hibernate + Oracle) serving 6M+ residential subscribers for account, billing, and appointment management.

Developed Java ETL applications processing 200K+ daily billing records from network provisioning, usage metering, and payment gateways into SQL Server for invoice generation.

Contributed to the migration of a Spring MVC monolith into Spring Boot microservices with Spring Data JPA, decomposing customer, billing, and service-activation modules.

Deployed applications on AWS EC2 with Chef configuration management, sustaining 99.5%+ availability across millions of daily customer transactions.

Optimized Hibernate queries and introduced Ehcache for reference data, reducing DB queries by 45% and page load times from 3s to under 1s.

Built Jenkins CI/CD workflows integrated with Gradle + GitHub, achieving 75%+ code coverage via automated JUnit tests.