Developed a scalable feedback analytics platform that consolidated and interpreted over 50,000+ reviews from diverse sources (web, app, social media), enabling cross-platform sentiment tracking. Built and deployed an asynchronous data scraping pipeline using Celery, Django, and Redis, achieving 3x faster processing. Leveraged deep neural networks (BERT, LSTM) for real-time NLP-based feedback classification. Automated feedback ingestion via Cron jobs and periodic data pulls, improving data freshness by 90%. Achieved 92% classification accuracy and 95% sentiment analysis accuracy using fine-tuned transformer models. Integrated a Gemini RAG-based chatbot to enable stakeholders to query actionable insights, enhancing decision-making efficiency by 60%.

Achieved 1st place in HackNUThon by developing AI-powered fraud detection middleware with millisecond response times, enabling real-time transaction monitoring while reducing fraud response time by 90%. Engineered comprehensive real-time alert dashboard with advanced analytics and risk scoring algorithms, providing contextual alerts with recommended actions and visualizing transaction patterns to improve fraud detection accuracy. Implemented RAG-powered AI chatbot assistant for instant transaction queries and compliance support, while integrating automated AML/KYC compliance features to ensure regulatory adherence without manual intervention. Built comprehensive policy management system with version control and historical transaction analysis capabilities.