SHARANYA BHATTACHARYYA

Motivated and detail-oriented Information Technology undergraduate at Kalinga Institute of Industrial Technology with a strong academic foundation and hands-on experience in machine learning, deep learning, and mobile application development. Proven ability to implement complex neural network architectures, demonstrated through a Summer Research Internship at IIT Guwahati and multiple advanced projects involving image classification, generative models, and Android development. Adept in C++, Python, Java, and Kotlin with a deep understanding of data structures, algorithms, and object-oriented programming. Passionate about research, open-source technologies, and building impactful, user-centric solutions.

Deep Learning Image Classifier using Triplet Loss Function | C++, CMake, Deep Learning | 2025

• Implemented sophisticated image classification system using triplet loss function achieving superior performance in learning discriminative embeddings with 128-dimensional representations.
• Developed custom C++ architecture for efficient neural network computation with L2 normalization techniques and semi-hard negative mining strategies.
• Fully Interactive Menu driven program including a command line interface (optional).

Mobile Gaming Widget Suite | Kotlin, XML | 2025

• Developed comprehensive collection of classic game widgets including Tetris (complete game logic with piece rotation and scoring), Tic-Tac-Toe (AI-powered opponent using Minimax algorithm), and 2048 (sliding puzzle with smooth mechanics).
• Utilized modern Android development practices and efficient memory management techniques.

Sorting Visualization Application | Kotlin, XML | 2025

• Created interactive mobile application to visualize sorting algorithms (Quick Sort, Merge Sort, Bubble Sort, Insertion Sort, Selection Sort) with real-time step-by-step animations.
• Implemented customizable speed controls, algorithm selection, and visual highlighting to enhance educational value and user engagement.

Generative AI Image Generator | Python, Machine Learning | 2024

· Developed advanced image generation model using Generative Adversarial Networks (GANs) with datasets from Kaggle and Hugging Face.

• Implemented sophisticated neural network architectures with convolutional layers, batch normalization, and progressive training strategies.

"Triplet Loss Function: Implementation, Features and Applications in Deep Learning-Image Classification and Face Recognition" (Under Review)
Authored during Summer Research Internship at IIT Guwahati

• Analysed semi-hard negative mining strategies and documented key advantages including superior similarity learning, direct embedding optimization, and open-set recognition capabilities.
• Provided practical implementation guidelines for optimal performance including learning rates (0.0003), margins (0.2), batch sizes (64-128), and mining strategies.