• Category: Web Design
• Project Date: Nov 2023
• Type: CSCI 571: Web Technologies

Description:

• Created a webpage that allows users to search for items on eBay using the eBay Finding API, displaying results in a tabular format.
• Used AJAX to send search queries from the webpage to a Flask-based Python backend, which will handle requests and responses in JSON format.
• Utilized HTML5, Bootstrap, and Angular for a responsive user interface, enhancing the user experience.
• Hosted the application on Google Cloud Platform (GCP) and used Node.js with Express for additional server-side operations.

• Category: App Design
• Project Date: Dec. 2023
• Type: CSCI 571: Web Technologies

Description:

• Learned Swift, Xcode, and iOS app development to build a visually appealing iOS mobile application.
• Implemented functionality for users to search products via eBay APIs, view product details, and interact with the data.
• Used MongoDB to store items in a wishlist, managed through a Node.js backend service.
• Included a feature to allow users to post selected products to Facebook.

• Category: AI
• Project Date: Mar. 2024
• Type: CSCI 561: Foundations of AI

Description:

• Created an AI agent that plays Duo-Othello on a 12x12 board by reading game states, selecting optimal moves, and writing the results.
• Learned to apply the minimax algorithm and design an evaluation function, while managing the agent's time for efficient decision-making.
• Utilized Python, Vocareum, and file I/O for handling game input/output and optimizing AI performance.
• Gained experience in game theory, AI search algorithms, and time management in competitive programming.

• Category: AI
• Project Date: Dec. 2021
• Type: Personal Project

Description:

• Created a 2-player Memory game where the user competes against an AI, using a grid-based card-matching system.
• Designed AI with a knowledge base to improve its card selection and maximize its chances of winning.
• Utilized python programming language and frameworks for game development.
• Gained experience in game development, AI decision-making, and the implementation of knowledge-based systems.

• Category: Deep Learning
• Project Date: Apr. 2022
• Type: Personal Project

Description:

• Implemented Optical Character Recognition (OCR) for Hindi, English, and Marathi using manual processing for Hindi and EasyOCR for English and Marathi.
• Converted extracted text into audio files using the gTTS Python library, creating mp3 outputs for text-to-speech functionality.
• Learned about OCR processes, text segmentation, localization, and how to integrate audio output for multiple languages.
• Used Python packages EasyOCR and gTTS for text extraction and speech conversion, with CNN for image classification in Devanagari OCR.

• Category: Image Classification
• Project Date: Dec. 2023
• Type: Personal Project

Description:

• Built and compared multiple deep learning models (ResNet50, VGG16, InceptionResNet, CNN) to classify skin lesions as benign or malignant using the SIIM-ISIC dataset.
• Addressed class imbalance with data augmentation, scaled images, and balanced datasets for effective training.
• Used SHAP to interpret ResNet50's decision-making, highlighting features that influence the model’s predictions.
• Used Keras for model building, Scikit-Learn for balancing the data, and SHAP for explainability analysis. Learned about model performance evaluation and AI transparency in medical diagnostics.

• Category: Neural Netorks
• Project Date: Nov. 2021
• Type: Personal Project

Description:

• Built an Artificial Neural Network (ANN) model to classify various denominations of currency, using a self-created dataset of bank-note images.
• Used Principal Component Analysis (PCA) to reduce data dimensionality and separate correlated features, improving model accuracy and speed.
• Compared ANN models with and without PCA, achieving 90% accuracy and significantly reduced loss with PCA.
• Applied Eigen Face for identifying distinguishing features between real and fake currency.

• Category: Neural Netorks
• Project Date: Nov. 2022
• Type: Personal Project

Description:

• Built a face-recognition-based entry system using web-scraped images, focusing on detecting, aligning, and recognizing faces for attendance automation.
• Implemented models like KNN, SVM, ANN, and VGG16. Applied PCA for dimensionality reduction, improving ANN accuracy from 30% to 56% and boosting SVM to 63%.
• Leveraged the pre-trained VGG16 model for image classification, achieving superior performance compared to other models.
• Learned how PCA improves model efficiency, explored different machine learning algorithms for face recognition, and applied VGG16 for enhanced accuracy.

• Category: Image Processing
• Project Date: Nov. 2021
• Type: Personal Project

Description:

• The aim was to remove noise from images by converting them to the frequency domain, reducing high-frequency noise, and then transforming them back to the spatial domain.
• Utilized Fourier Series and Fourier Transform concepts to decompose and analyze image frequencies, distinguishing between noise and image details.
• Converted images to frequency domain using Fourier Transform, filtered out high-frequency noise, and used Inverse Fourier Transform to reconstruct and restore the original image.
• Gained insights into frequency domain processing for image denoising and the importance of Fourier Transform in separating and mitigating noise from useful image information.

• Category: NLP
• Project Date: May 2022
• Type: OMDENA

Description:

• Collected medical data on six diseases and cleaned over 90,000 tweets, removing non-English content and duplicates.
• Used VADER to analyze and classify tweet sentiments into positive, negative, and neutral categories.
• Applied active learning for text classification to identify relevant (patient experience) and irrelevant tweets.
• Enhanced chatbot capabilities by incorporating sentiment analysis to better understand and address patient emotions and experiences.

• Category: NLP
• Project Date: Nov. 2022
• Type: Personal Project

Description:

• Developed a near real-time warning system for cyberbullying by classifying conversation risks into low, moderate, or high using machine learning and sentiment analysis.
• Consolidated and cleaned 807,000 messages from multiple datasets, including PAN12 and Mendeley, and preprocessed text for analysis using tokenization, lemmatization, stemming, POS tagging, and vectorization.
• Utilized VADER for initial sentiment analysis, employed various machine learning models (e.g., Stochastic Gradient Descent, Random Forest), and applied deep learning with a Deep Neural Network and fastText for superior classification accuracy.
• Achieved the highest accuracy of 94.70% with fastText. Gained experience in handling large datasets, implementing active learning for annotation, and applying advanced text classification techniques.

• Category: NLP
• Project Date: Apr. 2022
• Type: Personal Project

Description:

• Created a resume filtering application using Python and Streamlit to recommend top candidates based on job description similarity.
• Collected and cleaned resume and job description data, including basic text preprocessing tasks.
• Implemented similarity scoring and topic modeling to recommend the top 10 candidates with the highest similarity to job descriptions thus optimizing candidate selection based on relevance.
• Gained experience in text similarity measurement, topic modeling, and integrating these techniques into a user-friendly application for effective resume filtering.

• Category: Machine Learning
• Project Date: Jan. 2022
• Type: Personal Project

Description:

• Created a dashboard for monitoring and predicting water quality using satellite imagery. Integrated indices like NDWI, MNDWI, and NDTI to analyze parameters such as pH, turbidity, and dissolved oxygen.
• Collected and processed satellite data from Sentinel 2, Sentinel 3, and Landsat 8. Applied machine learning, including Random Forest Classifier, to predict water quality categories and handle imbalanced data with SMOTE.
• Built an interactive dashboard using visualizations to help decision-makers assess water quality in real-time. Implemented features for parameter selection, data visualization, and comparative analysis.
• Utilized Python for data extraction and analysis, Google Earth Engine for satellite data processing, and Streamlit for dashboard development. Employed machine learning models for classification and data balancing.

• Category: Machine Learning
• Project Date: Nov. 2021
• Type: Personal Project

Description:

• Applied various machine learning algorithms to classify anxiety levels from the Taylor Manifest Anxiety Scale (TMAS) dataset. Algorithms included Naïve Bayes, KNN, Logistic Regression, SVM, Decision Tree, and Random Forest, with techniques like GridSearchCV, K-fold cross-validation, Adaboost, Gradient Boosting, and Voting Classifier.
• The Voting Classifier achieved the highest accuracy (96.95%), followed closely by Gradient Boosting with SVM as the base estimator (96.67%). Important predictors of anxiety included specific questionnaire items related to self-perception and emotional responses.
• Gained hands-on experience in implementing and tuning various machine learning models for classification tasks. Understood the impact of boosting techniques and ensemble methods on model performance.
• Utilized GridSearchCV for hyperparameter tuning, K-fold cross-validation for model validation, and applied boosting techniques like Adaboost and Gradient Boosting to enhance classifier accuracy.

• Category: Machine Learning
• Project Date: Apr. 2021
• Type: Personal Project

Description:

• Built and evaluated a Linear Regression model to predict house prices based on various features. The process involved data cleaning, feature selection using Recursive Feature Elimination (RFE) and Backward Selection, and model evaluation.
• Gained expertise in data preprocessing and cleaning, performed feature selection to improve model accuracy, and applied diagnostic tests to validate model assumptions including normality, homoscedasticity, and multicollinearity.
• Utilized Linear Regression for prediction, RFE and Backward Selection for feature reduction, and conducted assumption tests such as Anderson-Darling, Shapiro-Wilk, Breusch-Pagan, and VIF. Visualized data using scatter plots, histograms, and Q-Q plots for model evaluation.
• The final model achieved high performance with an R-squared score of 0.918 on the training set and 0.915 on the test set, indicating strong predictive power.

• Category: DBMS
• Project Date: Apr. 2021
• Type: Personal Project

Description:

• Developed an online voting system using Python and MySQL to address the challenges of in-person voting during the pandemic. The system allows voters to cast their votes remotely, aiming to improve accessibility and accuracy.
• Gained experience in designing a relational database with MySQL, creating an Entity-Relationship Diagram (ERD) to model database relationships, and implementing an online voting interface with Python.
• Utilized MySQL for database management, Python for developing the online voting application, and ERD for visualizing the database schema and relationships among various entities.
• Designed tables for managing elections, parties, candidates, voters, and votes, ensuring a secure and accurate voting process with features for managing user information and vote tallying.

• Category: Data Analysis
• Project Date: Oct. 2021
• Type: Research Project

Description:

• Analyzed crop yield data for Telangana over three years (2016-2018) by examining factors such as temperature, rainfall, and crop types (Bengal gram, groundnut, and maize). The process included data cleaning, integration, visualization, PCA (Principal Component Analysis), and linear regression modeling to predict future crop yields.
• Gained insights into the impact of climatic factors on crop yields, and the significance of rainfall and temperature. Learned to use PCA for dimensionality reduction and identify key factors affecting crop yield. Utilized linear regression and optimization techniques for accurate predictions.
• Employed Python for data analysis and linear regression, Power BI for visualization, and PCA for factor analysis. The integration of various datasets provided a comprehensive view of crop yield influences.
• Suggested strategies for water conservation and rainwater harvesting to improve crop yields, with specific recommendations for urban, rural, and industrial areas to enhance water management and support sustainable agriculture in Telangana.

• Category: Statistics
• Project Date: Nov. 2020
• Type: Personal Project

Description:

• Analyzed the importance and preference for home remedies across different generations using a Naive Bayes classifier. Collected data manually through a survey to understand preferences for home remedies versus professional help for common health issues.
• Gained insights into the effectiveness of home remedies and their usage patterns across generations. Learned to apply the Naive Bayes theorem for classification tasks and to use hypothesis testing for statistical analysis.
• Employed Naive Bayes algorithm for classification with a model accuracy of 91.11%. Conducted data analysis to interpret survey results and used hypothesis testing to validate findings.
• The model accurately predicted preferences for home remedies, highlighting their significant role in health care. The analysis provided a clear understanding of home remedy usage trends across generations.