Manish Kumar Roy Email & Phone Number

Kolkata, Wb, In

Coventry, England, United Kingdom

Project Title: Real-time Electro-Thermo-Mechanical Performance of Laser Welded Joints in Li-ion Cells •Selected 1 mm thick Aluminum plate(1050) and used a waterjet cutting machine to shape specimens, and deburred edges with precision •Designed fixtures to securely hold specimens during laser welding, optimizing alignment and minimizing deformation during processing •Programmed and Successfully conducted laser welding on 3 different types of specimens: Linear, Circular, and Rectangular laser welding • Implemented three key variations in power settings to consistently achieve the desired weld depth suitable for industrial application •Performed tensile testing, measured voltage changes under constant 200A current, monitored thermal profiles using a thermal camera •Achieved peak load for Rectangular and Circular laser welding at approximately 1400N, double that of linear laser welding at 600N •Maintained resistance and temperature consistently similar across all types; breaking load followed the same trend, at 140N and 60N

Sunnyvale, California, United States

Project title: Automated News Recommendation System Using Twitter Activity Analysis •Developed a personalized news recommendation system suggesting relevant articles based on users' Twitter activity and reading history •Applied collaborative filtering to analyze Twitter users' reading history and content-based filtering to match article content with user tweet •Performed data collection and analysis, including sentiment analysis, topic modeling, user clustering to improve recommendation accuracy •Improved user engagement by 20% and significantly increased click-through rates by 15%, based on data from 10,000 user interaction

Kharagpur, West Bengal, India

•Developed a Deep learning model for heart sound classification, by integrating 832 audio files from the iPhone app and digital stethoscope •Used advanced preprocessing like volume normalization, noise reduction, audio segmentation, and MFCC for feature extraction •Engineered Bidirectional LSTM model, optimized layers, and hyperparameters for sequential audio analysis with 199,700+ parameters •Achieved an overall accuracy of 78%, indicating effective precision and recall across the category, showcasing the models' robustness

Iit Kharagpur

•Guided 4 students, ensuring a smooth transition to institute life and addressing diverse academic and extracurricular challenges with support •Conducted meetings, advising first-year students on academic and extracurricular decisions as part of the Dean's initiative at IIT Kharagpur

Project title: Advanced Transfer Learning Techniques for High-Accuracy Plant Disease Detection and Classification• Effectively resized and converted 5,000+ plant images (128x128 pixels) across 10 species, ensuring ~200 images per disease category• Successfully implemented ResNet, VGG16, VGG19, Inception, and ImageNet models, achieving up to 95% accuracy in disease classification• Significantly fine-tuned models with a 0.001 learning rate, 50% dropout, and strategic data augmentation, effectively reducing overfitting• Achieved remarkable accuracy of 90-95% with 2-hour model training time per model, significantly enhancing disease detection capabilities

•Led data analysis for Social Buzz, uncovering content categories and delivering the key strategic insights for optimized content strategies •Utilized SQL, Excel, and PowerPoint to meticulously clean and visualize data, presenting actionable insights for informed content creation

Aerospace Engineering, Dual Degree, 8.31

Minor: Computer Science & Engineering (B.Tech 4Y) Micro: Entrepreneurship & Innovation, Artificial Intelligence and Applications

Education record

Higher Secondary, Physics, Chemistry, Mathematics, English & Physical Education, 88.4%

Secondary, 9.8/10