SHUBHAM KUMAR

Accelerated User Acceptance Testing (UAT) for over 5 derived variables by interpreting JIRA documentation, enhancing AML model capabilities.

Optimized UK Merchant Collusion detection pipeline code, significantly improving processing efficiency for over 3 million customers.

Conducted in-depth data analysis using SQL and Python on large datasets in Hive and BigQuery to inform and refine financial crime AI models.

Applied advanced LLM techniques, including semantic and recursive chunking with LangChain, to process over 100 large documents, significantly improving information retrieval and response accuracy.

Rigorously tested a chatbot across 50+ documents, achieving high response accuracy for over 1,000 users and saving 10% in man-hours by enhancing OTP processes.

Developed multiple Power BI dashboards for critical business KPI visualization, reducing manual reporting efforts by over 10 hours per week.

Recognized as a 6e Fab Debutant (top 5 out of 100+ new joiners) for exemplary performance, awarded by the SVP.

Implemented the Brotli algorithm to compress TUDF files, achieving over 30% data compression without any loss.

Developed a proof-of-concept for file compression, demonstrating a 25-30% improvement in compression size compared to the previous algorithm.

Researched and developed a robust lossless file compression technique optimized for managing over 70 TB of data.