Aman Devrani

A software-based AI system leveraging YOLOv8 object detection and PyTorch deep learning to analyze live traffic video feeds in real-time. Dynamically optimizes traffic signal timings based on vehicle density and movement patterns, reducing idle time at intersections. Uses PySpark for processing large-scale traffic datasets and OpenCV for computer vision tasks. Deployed via Flask API for seamless integration with existing traffic infrastructure, promoting fuel efficiency and reducing carbon emissions.

Impact: Reduces vehicle idle time and CO₂ emissions through real-time AI-driven traffic signal optimization, deployable on existing infrastructure

An intelligent DBMS-based bus management system for real-time bus tracking, automated seat allocation, and scheduling with three portals for students, drivers and, admins. Features live geolocation tracking using Leaflet.js maps and WebSocket-based real-time communication, enabling students to monitor bus locations and drivers to update routes dynamically.

Impact: Enables real-time bus tracking and automated resource allocation, reducing operational inefficiencies and improving campus transport coordination

A containerization-based platform using Docker to provide isolated coding environments for educational institutions on minimal hardware. Features a React frontend for student interaction, Node.js backend for resource management, and Shell scripts for user and admin activities. Includes personalized learning analytics, and Prometheus/Grafana monitoring for system performance. Enables multiple students to work simultaneously on shared infrastructure with instant boot times.

Impact: 70% reduction in initial cost and 80% reduction in operational costs compared to traditional labs, with boot times reduced from minutes to seconds

A modern, responsive personal portfolio website built with Next.js 16 and TypeScript, featuring smooth animations powered by Framer Motion and elegant styling with Tailwind CSS. Showcases technical skills, professional experience, academic achievements, and project portfolio across Web Development, AI/ML, and Data Science. Includes interactive sections for skills visualization, detailed project cards, certifications display, and contact form. Optimized for performance with server-side rendering and static generation for fast load times and excellent SEO.

Impact: Serves as a comprehensive professional digital identity, enabling recruiters and collaborators to explore technical expertise through an interactive platform. Demonstrates proficiency in modern web development, UI/UX design, and performance optimization while facilitating networking opportunities and showcasing continuous learning.

An intelligent cloud resource allocation system implementing the 0/1 Knapsack dynamic programming algorithm to optimize virtual machine distribution across cloud infrastructure. Built with C++ for high-performance computation, Flask backend for API services, and React frontend for visualization and monitoring. Analyzes resource requirements and constraints to minimize VM wastage while maximizing utilization efficiency. Includes real-time allocation dashboard and performance metrics tracking for cloud administrators.

Impact: Enhanced performance and reliability of VM allocation

A machine learning application that predicts house prices based on geographical location and property attributes using advanced regression algorithms. Implements multiple models including Linear Regression, Random Forest, and Gradient Boosting with Scikit-learn for optimal accuracy. Features data preprocessing, feature engineering, and model comparison to select the best performing algorithm. Deployed as a Flask web application with an intuitive interface for users to input property details and receive instant price predictions.

Impact: Achieved 93% prediction accuracy on test dataset

A comprehensive machine learning system for pre-approved loan allocation decisions, comparing multiple classification algorithms including Logistic Regression, Decision Trees, Random Forest, and Neural Networks. Utilizes Pandas for data manipulation and NumPy for numerical computations on large financial datasets. Implements feature selection, cross-validation, and hyperparameter tuning to optimize model performance. Provides detailed analysis of applicant creditworthiness with explainable AI techniques for transparent decision-making in financial services.

Impact: Achieved 95% accuracy in loan approval predictions