NagaKishoreRaju D

Merchant centralized data operations -

Merchant Organized Data layer are physical views of frequently used merchant data. Merchant ODL tables contain raw data from system of records or derived variables to create consistent and easy access to frequently required merchant information by multiple stakeholders for decisioning, reporting and analytical needs.

Performance data stabilization -

Performance data stabilization data is used to generate the monthly global merchant results and end of month global rolling 36 month financial reports. This data is used across multiple regions.

Key responsibilities -

• Contributed to requirement gathering, developing and deploying solutions that meet both technical and functional requirements.
• Accelerated high-computation workflows by migrating code from Hive to PySpark, achieving significant reductions in processing time.
• Developed and implemented sophisticated business logic solutions in analytical processing, driving insights and data-driven decision-making.
• Implemented data pipelines with best practices, incorporating robust quality checks, validations, and optimization techniques to ensure data integrity and efficiency.
• Created comprehensive unit test cases to rigorously test code module scenarios, ensuring robust functionality and reliability.
• Built high-performance applications leveraging in-memory computing technology to achieve rapid processing and scalability.
• Enhanced the performance of PySpark and Hive codebase through targeted tuning and optimization techniques
• Applied optimization techniques to boost script speed and efficiency.
• Achieved a 50% reduction in end-to-end processing time, decreasing execution time from 17 hours to 8 hours.
• Participated in the end-to-end software development lifecycle, encompassing building, testing, and deploying high-quality code.
• Successfully designed and developed a comprehensive project from concept to delivery.

Advance Emergency Breaking Systems-

Advanced emergency braking systems utilize sensors to continuously scan the road and detect potential collisions. The system detects potential collisions and responds by alerting the driver and, if required, autonomously applying the vehicle's brakes to prevent a collision or reduce its severity. The core goal is to build an application that utilizes newly generated onboard data from DACU modules to train machine learning models, enabling insights into errors and improvements in advanced emergency braking systems. The project is divided into three interconnected segments: data pipeline creation, Analytics layer development, and Reporting layer implementation, with the Reporting layer receiving insights from the Analytics layer.

Key Responsibilities -

• Designed and implemented a data migration pipeline to integrate legacy data into Hadoop, facilitating data modernization and business insights.
• Built SQOOP scripts to migrate data from SQL Server to Hadoop, leveraging big data capabilities for advanced analytics.
• Utilized PySpark and native Python scripts to perform data transformations during the ETL phase, ensuring data quality and consistency.
• Designed and orchestrated data pipelines with Jenkins, leveraging automated configurations for efficient data management.
• Created BigSQL scripts to extract data and developed a data ingestion component, enabling machine learning lifecycle processes.
• Developed a Python framework to automate and schedule advanced machine learning analytics workflows, enabling efficient and scalable insights generation.
• Architected a database layer with integrated job-level auditing, supporting a scalable 3-layered application architecture.
• Designed scripts for data quality assurance, implementing versioning in the historical layer to guarantee data consistency and reliability.
• Unified the reporting layer with the reporting tool, delivering data-driven insights and visualizations to users.

ETooling Image Recognition -

The primary objective of this project is to automate decision-making processes by leveraging Artificial Intelligence (AI) solutions, eliminating the need for manual judgment. The existing process relies on human analysis of image sets, with subsequent calculation and application of business rules to derive insights. The newly developed component will automate end-to-end processes, eliminating human intervention and adding significant value to the existing business.

Key responsibilities -

• Integrated computer vision API calls to extract text from inputs, applying cognitive models to build knowledge and enable intelligent insights.
• Developed a data extraction component that taps into Azure Storage's Blob Container and Table storage capabilities, streamlining data extraction and processing.
• Creating data transformations aligned with business rules to enhance decision-making processes.
• Created a robust testing framework using PostgreSQL DB, facilitating automated testing and reducing manual effort.
• Reporting data will be transformed into engaging and informative data visualizations, enabling data-driven decision-making.
• A responsive UI was built using Django, providing a seamless user experience for file uploads and batch initiation.

Diagnosis extract Solution-

The DX Volatility Report offers valuable insights into Claims data instability for administered drugs, providing monthly evaluations at procedure code and provider levels.

Key responsibilities -

• Weekly Dx data ingestion from Oracle to Hadoop is automated using Sqoop, ensuring a consistent and reliable data flow.
• To enable data analysis, metadata was established in the Hive environment, transforming ingested data into valuable insights.
• Reports are dynamically generated using Spark SQL, offering flexible filtering options such as time periods, top providers, and custom business calculations.
• Quality checks are implemented using Hive, employing statistical elements like moving averages and central tendency to monitor and ensure data quality.
• Experienced in efficiently managing big data using Spark's capabilities, including partitioning, in-memory processing, broadcasts, and effective join and transformation techniques.
• Developed custom code to integrate report storage into the HDFS environment, ensuring secure and scalable data management.
• PySpark was utilized to design and develop high-performance custom code, unlocking efficient data processing capabilities.

Risk Reporting-

The project's objective is to design a data loading process for credit migration tables, utilizing a CSL to enable flexible data distribution to various analytical applications within the semantic layer architecture.

Key responsibilities -

• Spark was utilized to design a data ingestion pipeline, loading data from curated and DAL zones into a common staging area.
• To ensure transparency and accountability, auditing tables capture detailed job execution steps.
• Metadata tables were designed to serve as the foundation for the semantic layer, driving data meaning and context.
• Implemented stored procedures to consolidate semantic layer data, enabling reporting and analytics capabilities.
• The data ingestion process was enhanced with the implementation of Sqoop, ensuring reliable and high-performance data transfer.
• Oozie workflows were leveraged to coordinate and schedule tasks, providing a robust workflow management solution.