Praveen Yeduresi

– Designed and implemented an end-to-end fraud detection analytics platform to monitor transactional and behavioural data across multiple client accounts. The platform combined rule-based and predictive analytics to identify high-risk activity and emerging fraud trends while ensuring compliance with GDPR and PCI DSS regulations. – Consolidated transactional, behavioural, and customer data from multiple sources for unified analysis, enabling faster insights. – Applied rule-based logic and predictive models to detect high-risk transactions, increasing early fraud detection by 35%. – Developed interactive dashboards in Power BI and Tableau for real-time monitoring and executive reporting. – Reduced manual fraud investigation effort by 40%, streamlining operations across multiple client accounts. – Delivered actionable insights for fraud mitigation while maintaining regulatory compliance and minimal disruption to legitimate customers.

– Developed a scalable, research-focused Q&A system leveraging Retrieval-Augmented Generation (RAG) for accurate, context-aware responses. Integrated Gemini Pro and LangChain to enable efficient access to a large corpus of research papers and knowledge sources, supporting real-time, high-quality information retrieval and dissemination. – Consolidated research papers, technical documents, and knowledge repositories into a unified, indexed knowledge base for efficient retrieval. – Applied NLP techniques for document embedding and vector-based similarity search, enabling rapid retrieval of relevant research data. – Developed and implemented a scalable, real-time RAG pipeline supporting thousands of concurrent user queries. – Validated system performance through experiments, ensuring high response quality and user engagement.

– Collaborated with academic and IT stakeholders to define KPIs for performance monitoring and compliance reporting. – Built Databricks data pipelines to clean, transform and integrate data from admissions, attendance and exam datasets. – Automated data preparation workflows, reducing manual effort by approximately 35%. – Developed Power BI dashboards for student enrolment trends, faculty workload and resource utilisation. – Reduced reporting turnaround time by around 40%, enabling faster operational decision-making.

– Led a data analytics initiative to identify customer behaviour patterns and predict churn, enabling targeted interventions and improved operational planning. – Cleaned and transformed large datasets using Excel, SQL, and Python, improving data reliability. – Conducted trend analysis and segmentation to identify key drivers of customer behaviour. – Built predictive models that reduced customer churn by 15%, supporting revenue growth. – Identified data inconsistencies and implemented validation checks to improve data accuracy. – Developed dashboards and reports to present insights in a clear, concise format. – Worked closely with stakeholders to align analysis with business needs and improve decision-making. – Recommended process improvements based on data insights, enhancing operational efficiency. – Developed analytical dashboards to provide procurement teams with visibility into historical market price trends and seasonal patterns for key horticultural procurement inputs (including fertilisers, seeds and crop-protection inputs), while preparing clean, structured datasets to support price forecasting and proactive purchasing decisions. – Ingested and prepared historical procurement input price and seasonal datasets using Azure SQL and Python. – Performed trend, variance and seasonality analysis to identify procurement cycles, price peaks and dips. – Built Power BI dashboards showing monthly price trends, category-level patterns and forecast ranges. – Enabled earlier visibility into input price fluctuations, helping reduce procurement cost risk by approximately 10-12% and improve budget planning. – Collaborated with the data science team by providing high-quality, well-structured datasets, contributing to a 15% improvement in forecast accuracy.