KISHAIYAN VELLAICHAMY THANGARAJ

Strategic Software Engineer skilled in application development, testing and optimization. Excels at coordinating ground-up planning, programming, and implementation for core modules. Maintains strong object-oriented and software architecture fundamentals.

• Certification in Java
• Amazon Web Services Certified Developer
• Certification in C++, Mobile Application Development and Machine Learning

• Best Outgoing Student 2015
• Owning Patent on Public Toilet Monitoring Sensor

CropSense Application Development:

• Flutter used to create a user-friendly and responsive chat application. Integration with ChatGPT for natural language understanding and generation.
• Users can capture images of plant leaves within the chat application. Utilizes the Object Detection Model to identify and highlight the plant leaves in the image.
• Integration of a Segmentation Model for precise classification of leaf diseases. Users receive instant feedback on the detected disease through the chat interface.
• Natural language queries about plant health can be posed to ChatGPT.Users receive informative responses based on the integrated models' outputs.
• Additional feature providing information and suggestions for disease management. Rich media content and links shared through the chat interface for enhanced user education.
• Tech-stack:

       Flutter- Cross-platform Development

       Python Flask - for api calls

       Firebase - for user authentication and storage 

       TensorFlow- Training and exporting the models to the application

Electronic Voting System:

• Developed with the Microsoft-endorsed Secured by Design architecture, the Electronic Voting System has undergone a comprehensive development cycle.
• Implemented dual views to cater to distinct user roles, featuring interfaces tailored for both voters and administrators overseeing the entire election process.
• Integrated robust logging mechanisms to track and monitor the actions performed by the administrator. This log functionality ensures transparency and accountability, allowing for a detailed record of the admin's activities throughout the election process.
• Incorporated Multi-Factor Authentication (MFA) integration to bolster security measures. Every user, including both voters and administrators, is required to enroll in MFA, ensuring a robust authentication process during login and vote casting. This added layer of security enhances the overall integrity and confidentiality of the Electronic Voting System.
• The administrator assumes a pivotal role, managing the setup of the election and vigilantly monitoring for any anomalies or irregularities throughout the entire electoral process.
• Tech-stack

       React.js- for Front-End Development

       Firebase- for user authentication,MFA,Storage,and rules for more Security

       Source and Version Control: Git, GitHub

Recommender System using Association rule/pattern mining:

• Our Recommender System merges the Apriori Algorithm for Association Rule Mining with Collaborative Filtering, offering a comprehensive solution for personalized recommendations.
• Leveraging both historical user interactions and frequent itemsets discovered by Apriori, the system provides context-aware recommendations, ensuring a tailored user experience.
• Recommendations are intelligently ranked using a combination of collaborative filtering scores and Apriori-generated association strengths, optimizing relevance for users.
• The project seamlessly combines Memory-Based Collaborative Filtering, User-Based Model Collaborative Filtering, and Apriori Algorithm, resulting in a robust and versatile hybrid recommendation engine.
• By prioritizing user preferences and uncovering novel items through association patterns, the Recommender System elevates user satisfaction and engagement, contributing to an enriched platform experience.
• Tech-stack: Python, Jupyter Notebooks

Detection of Brain Tumour:

• CT scan was imported as DCM file which will undergo preprocessing like resizing, image enhancement with CLAHE method.
• Then used multi-atlas segmentation to identify different layers.
• For dimensionality reduction, feature extraction algorithms PCA and LDA was used to optimize the features selected.
• Non-kernel SVM was used to predict the exact position of the tumor in the brain.
• Tech-stack: MatLab

Automated System for Monitoring Usage of Toilets Using IOT and Data Analytics

• Arduino UNO in combination of sensors is used to sense the change in the ammonia level or Hydrogen Sulphide which correspond to the use of toilet is programmed in C++ and is used to update the count of people used and number times cleaned in a website written in HTML, CSS and JavaScript.
• The data is then used to analyze where the people use the toilet and demands more people to maintain.
• Tech-stack: Arduino , Microsoft Azure