Brijesh Kumar Morya

Problem: Existing expense tracking apps are often complex, lack real-time collaboration, and do not support UPI or flexible split methods suitable for Indian users. Solution: Developed a full-stack expense management application that enables users to track personal and group expenses, split bills using equal, percentage, or custom methods, and settle dues through UPI/QR-based payments. The system maintains real-time balances and computes optimized net settlements to minimize transactions. Result: Simplified shared expense tracking with accurate calculations, faster settlements, and an improved user experience.

• Personal and group expense tracking with multiple split methods
• Friend and group management system
• UPI and QR-based payment settlement with verification flow
• Optimized net settlement algorithm to reduce transactions
• Real-time updates for expenses and settlements

Problem: Travel platforms need intuitive browsing, detailed listings, and secure user access to allow users to explore destinations and property information easily. Solution: Built a full-stack travel listings web application that supports user login, browseable destination listings, and individual listing detail views. The platform features secure session-based authentication, responsive UI, and efficient backend APIs. Result: Users can log in, view travel listings, and navigate to detailed property pages with consistent performance and smooth navigation.

• Secure login and user sessions using Passport.js
• Dynamic listings index with destination previews
• Individual property/detail views via dynamic route parameters
• RESTful backend APIs supporting listing data and user sessions
• Image uploads and optimization via Cloudinary

Problem: Traditional grievance handling in colleges is slow and manual, often taking weeks to route complaints to the correct authority. Solution: Built a system where students can submit grievances through text or voice input, with multilingual support. Submitted grievances are automatically classified and routed to the concerned authority, who reviews, resolves the issue, and updates the status. Result: Faster grievance routing, improved transparency, and reduced manual intervention in the complaint handling process.

• Text and voice-based grievance submission
• Multilingual support using language models
• Automatic grievance routing to concerned authority
• Status tracking for submitted complaints
• Collaborated with team members on UI and system integration

Problem: Text classification requires careful feature engineering and model selection. Solution: Built TF-IDF vectorizer (80K vocab) and ensemble pipeline with Gradient Boosting, Random Forest, and Logistic Regression on Kaggle's 44K article dataset. Optimized via grid search and cross-validation. Result: 99.48% accuracy, <1s CPU inference time.

• TF-IDF feature extraction (80K vocabulary)
• Gradient Boosting (99.48% accuracy)
• Random Forest & Logistic Regression ensemble
• 44K balanced Kaggle dataset
• Precision, Recall, F1-score evaluation