Balu Upendar Cholleti

Built a high-volume payout platform using Java, Spring Boot, REST APIs, and microservices to support payout initiation, transaction tracking, reconciliation, and exception recovery for millions of records daily. Developed Next.js/React/TypeScript operations dashboards to improve visibility into payout status and processing exceptions. Optimized payout workflows by using Redux Toolkit and React Query (including caching and retry strategies), reducing duplicate API calls by ~25%. Improved concurrent processing using Spring WebFlux and CompletableFuture to reduce peak-period latency by ~30%. Implemented fault-tolerant Spring Batch jobs with partitioning, checkpointing, retries, and skip handling for resumable large workloads. Built event-driven audit and reconciliation flows using Kafka, Google Cloud Pub/Sub, and Cloud Functions, reducing manual recovery effort by ~35%. Secured payment APIs with Spring Security, OAuth 2.0, JWT, and RBAC with audit logging. Automated Dockerized deployments

Contributed to an AI-enabled financial crime analytics application using Python and SQL by integrating customer, transaction, and AML alert data to improve investigation efficiency by 30%. Built relational data models and reusable transformation workflows to connect entities (customers, accounts, merchants, transactions), standardizing attributes and risk indicators and improving analysis-ready data availability by 30%. Developed Python/Pandas/SQL processing to transform and enrich 10M+ monthly transaction records, reducing manual preparation effort by ~35%, and automated validation checks for missing attributes, duplicates, and mapping inconsistencies. Engineered risk-related features and integrated Scikit-learn anomaly detection and risk scoring outputs into alert prioritization, improving high-risk case identification by 20%. Evaluated model performance using precision/recall and addressed false-positive analysis with compliance partners. Prototyped BERT-based NLP workflows with Hug