Aug 2023 – Present · 2 yrs 7 mos

• Designed and implemented end-to-end Data Collection and Distribution framework in Python using Apache Beam and Kafka, enabling real-time ingestion, transformation and publication of high-volume streaming data; reduced on-boarding time for new sources by 80% through metadata-driven configuration and ensured reliability via fault-tolerant design, logging, and cloud extensibility (Google Pub/Sub, BigQuery). Framework has been adopted by 5+ teams across the Contact Center domain as the foundation for ’ETL as a Service’ for the enterprise.
• Automation Software: Developed a Python-based XML parsing tool for streamlining the analysis of 2000+ SSIS packages and SQL jobs for end-to-end data lineage tracking, reducing manual analysis time by 90% and producing a structured documentation, thus accelerating transitioning to new contact center solution Verint and Five9.
• Built an Enterprise Pipeline Solution (EPL) in Jenkins to enable version-controlled SQL database deployments using Liquibase, automating schema changes across development, testing, and production environments. Integrated changelogs with Git and CI/CD workflows to ensure consistent, and auditable database updates, significantly reducing manual intervention, deployment risks, and enabling faster, compliant release cycles.
• Database Optimization: Facilitated data rationalization and scope isolation by identifying only application-relevant datasets from the broader upstream sources, eliminating redundant data ingestion, applying strategic indexing —resulting in a 30% improvement in production database query performance and significantly reducing long-term maintenance overhead.
• Modernization and maintenance of BankerPerformance, a .NET Angular-based legacy system. Responsibilities included library version management, ensuring unit and integration test coverage, addressing production defects to maintain system reliability and compliance.

Jul 2022 – Aug 2023 · 1 yr 1 mo

• Spearheaded the development of 30+ dynamic Tableau dashboards and UI forms to replace direct SQL-based employee data handling, transforming a technical bottleneck into a user-friendly, self-service interface.
• Designed role-based access experiences for supervisors, team leads, and managers, enabling secure viewing and modification of employee data (hierarchy, productivity, training, evaluations) without SQL knowledge or entitlements.
• Collaborated directly with business stakeholders to gather requirements in an ad-hoc, agile manner, iteratively delivering UI components aligned to organizational workflows and compliance needs, thus reducing turnaround time for data updates, and improving managerial productivity in daily operations.

Jan 2022 – Jul 2022 · 6 mos

• Implemented microfrontend architecture using Webpack 5’s Module Federation plugin for a ReactJS application, reducing load times to <50 ms.
• Designed runtime integration strategy for independently deployed microfrontends to communicate and share dependencies dynamically.
• Streamlined development workflows, fostering team independence and enabling vertical scaling of the application.

May 2021 – Jul 2021 · 2 mos · India · Remote

• Developed a near real-time fraud detection system capable of processing over 5 million daily transactions, achieving prediction latency under seconds for high-velocity data streams.
• Designed a multi-model detection pipeline combining:
• A rule-based engine for immediate elimination of obvious frauds.
• A transaction amount-based model using Isolation Forest.
• A location-based anomaly model using Local Outlier Factor, enabling detection of geographically inconsistent transactions.
• Trained ML models on 10 lakh+ historical transactions across ~1000 customers, improving accuracy and significantly reducing false positives by 10% compared to traditional time-series approaches.
• Implemented a streaming architecture using Kafka and PySpark, simulating realistic peak traffic (up to 500 transactions/sec) for robust model evaluation and load testing.
• Enabled adaptive learning by storing all transaction data in a database for periodic model retraining, aligning fraud detection with evolving user behavior.
• Enhanced customer experience and operational efficiency by reducing false complaints through context-aware, unsupervised anomaly detection strategies.