Sentiment Analysis on Native and Romanized Assamese Social Media Text
Built sentiment classifiers for native and Romanized Assamese text using Logistic Regression, Random Forest, LSTM, BiLSTM, and GRU.
Applied text preprocessing and feature extraction with NLTK and TF-IDF. Achieved ~70% accuracy, with deep learning models performing better on native text.
Python TensorFlow Scikit-learn Pandas NumPy
AI-Driven Real-Time Cloud Security Threat Detection System
Developed an AI-driven real-time cloud security threat detection system to monitor cloud infrastructure activity continuously.
Collected and processed AWS CloudTrail logs, storing structured events for analysis and historical tracking.
Implemented Isolation Forest–based anomaly detection to identify suspicious behaviors such as unusual logins, abnormal API calls, and bulk resource access.
Integrated an automated alert system to notify administrators of potential security threats in near real time.
Python AWS Scikit-learn
Smart Classroom Automation and Energy Monitoring System using Raspberry Pi
Developed a Raspberry Pi–based smart classroom automation system using PIR, IR, LDR, BME280, and energy monitoring sensors to automatically control lights and fans via relay modules.
Designed a cost-effective and scalable solution with support for future React Native–based remote monitoring, suitable for smart classroom and smart campus applications.
Python React Native TypeScript
Astitva Foundation Website
A website for an NGO to help the underprivileged and needy people.