My Projects
A collection of my most impactful work in software development and AI
Cat Name Genius
AI-powered application that analyzes cat photos and generates personalized name suggestions based on visual features.
Intelligence Report Generator
Automated system that transforms raw news data into concise, actionable intelligence reports using advanced NLP techniques.
Cryptography Tool
Interactive web application for exploring various cryptographic algorithms with a secure and user-friendly interface.
Password Strength Manager
A comprehensive password manager with real-time strength analysis, secure generation, and encrypted local storage.
Algorithm Visualizer
Interactive tool for visualizing sorting algorithms, comparing their performance, and learning through hands-on practice.
Distributed File System
A modern, scalable distributed file system with automatic replication, storage tiering, and data deduplication.
File Organizer
A Python utility that organizes files by various criteria such as type, date, size, or custom rules.
Alzheimer's Transcriptomics Analysis
Comprehensive transcriptomic analysis pipeline for Alzheimer's disease using RNA-seq data from temporal cortex samples.
Cat Name Genius
Project Overview
Cat Name Genius is an innovative web application that combines computer vision with creative naming algorithms. Using a sophisticated convolutional neural network trained on 50,000+ cat images, it analyzes feline features to generate personalized name suggestions.
Technical Highlights
- Implemented a CNN architecture using TensorFlow for accurate cat feature recognition
- Developed a custom name generation algorithm based on visual attributes
- Built a responsive web interface for seamless user interaction
- Achieved 92% user satisfaction rate in beta testing
Technical Architecture
Challenges & Solutions
Training Data Quality
Ensuring diverse and representative cat images across breeds, colors, and poses.
Solution:
Implemented data augmentation techniques and manual curation to improve dataset balance.
Model Optimization
Balancing model complexity with performance requirements for web deployment.
Solution:
Used model quantization and TensorFlow.js optimization to reduce model size by 60% while maintaining accuracy.
Intelligence Report Generator
Project Overview
An automated system that transforms raw news data into concise, actionable intelligence reports. This project showcases advanced natural language processing techniques and automated data analysis capabilities.
Technical Highlights
- Developed robust web scraping systems for multiple news sources
- Implemented advanced NLP algorithms for text summarization and analysis
- Created an automated scheduling system for regular report generation
- Integrated machine learning models for content relevance scoring
Technical Architecture
Challenges & Solutions
Source Variability
Handling the diverse formatting and structure of different news sources.
Solution:
Developed adaptive scraping modules with source-specific extraction rules and a unified data model.
Information Relevance
Distinguishing important information from noise in large volumes of text.
Solution:
Implemented a relevance scoring system using supervised machine learning with human-labeled training data.
Cryptography Tool
Project Overview
A comprehensive web application for exploring various cryptographic algorithms, including AES, RSA, and Vigenère Cipher, with a secure and interactive interface. This tool serves as both an educational resource and a practical utility for encryption and decryption operations.
Technical Highlights
- Implemented AES and RSA encryption for secure data handling with 256-bit key strength
- Developed a responsive React-based interface with real-time encryption visualization
- Built FastAPI backend with asynchronous request handling for optimal performance
- Achieved 95% test coverage with Pytest for security-critical encryption functions
- Implemented educational mode with step-by-step algorithm visualization
Technical Architecture
Security Measures
- Secure key generation with cryptographically secure random number generators
- Memory-safe implementation with secure erasure of sensitive data
- Rate limiting and input validation to prevent DoS attacks
Challenges & Solutions
Security vs. Performance
Balancing strong encryption with acceptable performance, especially for large files and browser-based encryption.
Solution:
Implemented progressive chunk-based encryption for large files and WebWorkers for non-blocking UI during intensive operations.
Algorithm Visualization
Creating clear, educational visualizations of complex cryptographic algorithms without exposing security vulnerabilities.
Solution:
Developed an abstract visualization layer that demonstrates algorithm concepts without revealing actual implementation details or security-critical parts.
Cross-Platform Key Exchange
Ensuring secure key exchange between frontend and backend while maintaining compatibility across platforms.
Solution:
Implemented a custom hybrid encryption scheme for key exchange, combining asymmetric encryption for initial exchange and symmetric for ongoing secure communication.
Password Strength Manager
Project Overview
A comprehensive password management application with real-time strength analysis, secure password generation, and encrypted local storage. This tool helps users create and maintain strong, unique passwords while understanding the factors that contribute to password security.
Technical Highlights
- Implemented sophisticated password strength analysis using zxcvbn with contextual feedback
- Developed entropy-based password generation with configurable parameters for length and character sets
- Created secure local storage with AES-256 encryption using CryptoJS
- Built with Vue 3 Composition API and TypeScript for type safety and maintainability
- Designed responsive UI with Tailwind CSS for seamless mobile and desktop experience
Technical Architecture
Security Features
- Master password never stored, only hash used for encryption key derivation
- Automatic clipboard clearing after 30 seconds
- Auto-logout after configurable period of inactivity
Challenges & Solutions
Local Storage Security
Securing sensitive password data in browser local storage without a server component.
Solution:
Implemented a multi-layered encryption approach with PBKDF2 key derivation and AES-256 encryption, ensuring data remains secure even if local storage is compromised.
Real-time Password Feedback
Providing meaningful, real-time feedback on password strength without overwhelming users.
Solution:
Created a custom scoring system with contextual, actionable feedback using zxcvbn analysis, featuring visual cues and specific suggestions for improvement.
Type Safety with Encryption
Maintaining TypeScript type safety when working with encrypted data and external encryption libraries.
Solution:
Developed custom type definitions and wrapper classes around CryptoJS functionality, ensuring type safety throughout the application while leveraging the encryption library's capabilities.
Algorithm Visualizer
Project Overview
An interactive educational tool for visualizing sorting algorithms, comparing their performance metrics, and learning algorithm fundamentals through hands-on practice. The visualizer helps users understand the inner workings of different algorithms through animated step-by-step execution.
Technical Highlights
- Visualized multiple sorting algorithms including Quick Sort, Merge Sort, Bubble Sort, and Heap Sort
- Created an interactive practice mode allowing users to manually step through algorithms
- Implemented real-time performance metrics tracking comparison counts, swaps, and execution time
- Designed responsive UI with customization options for array size, speed, and data distribution
- Developed an algorithm comparison mode to visualize multiple algorithms side-by-side
Technical Architecture
Implemented Algorithms
Challenges & Solutions
Animation Performance
Maintaining smooth animations and responsive UI during algorithm execution, especially with large data sets.
Solution:
Implemented a custom rendering engine with requestAnimationFrame and Web Workers for background processing, separating UI updates from algorithm execution.
Algorithm Step Tracking
Tracking and visualizing individual steps of complex recursive algorithms like Quick Sort and Merge Sort.
Solution:
Developed a step recording system that captures the algorithm's state at each operation, allowing for forward/backward navigation and playback without re-running the algorithm.
Educational Value
Balancing educational content with visual appeal while maintaining accuracy of algorithm implementation.
Solution:
Created an integrated learning mode with contextual explanations that highlight key algorithm concepts during execution, with time complexity analysis and best/worst case scenario demonstrations.
Distributed File System
Project Overview
A modern, scalable Distributed File System implementation featuring automatic file replication, storage tiering, and intelligent data deduplication. Built with Python (Flask) and React/Next.js, this system demonstrates enterprise-level architectural patterns and advanced data management capabilities.
Technical Highlights
- Implemented multi-tiered storage architecture with hot, warm, and cold tiers for optimal data management
- Developed automatic data replication and self-healing capabilities for high availability
- Created advanced data deduplication system to optimize storage usage
- Built file chunking service for efficient large file handling and parallel transfers
- Implemented encryption at rest using PBKDF2 and Fernet for data security
- Designed real-time system monitoring and node health detection
Technical Architecture
System Components
- Statistical Testing with Multiple Testing Correction
- Cell Type Specificity Mapping
- GO and Reactome Pathway Analysis
- Advanced Data Visualization
Challenges & Solutions
Data Heterogeneity
Handling biological and technical variability in transcriptomic data from different samples and donors.
Solution:
Implemented robust batch correction methods and incorporated donor-specific covariates in statistical models to account for variability while preserving biological signal.
Multiple Testing Burden
Managing the high false discovery rate when testing thousands of genes simultaneously.
Solution:
Applied advanced multiple testing correction methods including Benjamini-Hochberg FDR and evaluated p-value distributions to ensure statistical rigor.
Cell Type Deconvolution
Determining cell type-specific expression patterns from bulk tissue RNA-seq data.
Solution:
Developed a computational approach to map differentially expressed genes to cell type-specific expression databases, enabling identification of cell populations most affected in Alzheimer's.
Integration of Analysis Tools
Combining Python and R-based tools into a cohesive analysis pipeline.
Solution:
Created a seamless integration framework using rpy2, with careful management of data structures and memory to enable efficient data transfer between environments.
File Organizer
Project Overview
A Python-based utility that organizes files in a directory by various criteria such as file type, creation date, file size, or custom rules. This tool is designed for convenience and can handle a wide variety of file organization needs with a simple, modular structure.
Technical Highlights
- Implemented multiple organization strategies including by file type, date, size, and custom rules
- Developed a comprehensive logging system to track all file operations
- Created an extensible modular architecture for easy addition of new organization methods
- Built with cross-platform compatibility for Windows, macOS, and Linux
- Incorporated robust error handling for edge cases like file permission issues and duplicates
Technical Architecture
Project Structure
file-organizer/
├── main.py # Main script to run the organizer
├── organizers/ # Modules for different organization methods
│ ├── __init__.py
│ ├── by_file_type.py
│ ├── by_date.py
│ ├── by_size.py
│ └── custom_rules.py
├── utils/ # Utility functions and logging
│ ├── __init__.py
│ ├── file_utils.py
│ └── logger.py
├── logs/ # Log files
│ └── organizer.log
├── tests/ # Unit tests for each module
│ ├── __init__.py
│ ├── test_by_file_type.py
│ ├── test_by_date.py
│ └── test_by_size.py
├── LICENSE.md # License information
└── README.md # Project documentation
Challenges & Solutions
Cross-Platform Compatibility
Ensuring consistent behavior across different operating systems with varying file systems and path handling.
Solution:
Used Python's pathlib library instead of string manipulation for path handling, and implemented OS detection to adjust behavior for Windows, macOS, and Linux specific requirements.
File Collision Management
Handling duplicate filenames when organizing files into categorized folders.
Solution:
Implemented an intelligent renaming strategy with configurable options for appending incremental counters, timestamps, or custom prefixes to ensure no file data is lost during organization.
Performance with Large Directories
Maintaining acceptable performance when organizing directories with thousands of files.
Solution:
Developed lazy loading and batch processing mechanisms to reduce memory usage, along with progress tracking for large operations. Implemented multiprocessing for parallel file operations on suitable platforms.
Custom Rule Flexibility
Creating a simple yet powerful system for users to define custom organization rules.
Solution:
Designed a rule specification DSL (Domain Specific Language) that allows users to create complex rule sets using a simple syntax, with support for regular expressions and compound conditions.
Alzheimer's Transcriptomics Analysis
Project Overview
A comprehensive computational pipeline for analyzing gene expression in Alzheimer's disease using RNA-seq data from temporal cortex samples. This project integrates differential expression analysis, cell type specificity mapping, and pathway enrichment to identify molecular mechanisms underlying neurodegeneration.
Technical Highlights
- Implemented rigorous statistical testing framework for differential gene expression analysis
- Developed custom algorithms for mapping gene expression to specific neural cell types
- Created comprehensive Gene Set Enrichment Analysis (GSEA) workflow for pathway identification
- Built sophisticated data visualization tools for complex biological datasets
- Integrated Python and R analysis environments using rpy2 for seamless workflow
Technical Architecture
Analysis Workflow Components
- Data Normalization and Quality Control
- Statistical Testing with Multiple Testing Correction
- Cell Type Specificity Mapping
- GO and Reactome Pathway Analysis
- Advanced Data Visualization
Challenges & Solutions
Data Heterogeneity
Handling biological and technical variability in transcriptomic data from different samples and donors.
Solution:
Implemented robust batch correction methods and incorporated donor-specific covariates in statistical models to account for variability while preserving biological signal.
Multiple Testing Burden
Managing the high false discovery rate when testing thousands of genes simultaneously.
Solution:
Applied advanced multiple testing correction methods including Benjamini-Hochberg FDR and evaluated p-value distributions to ensure statistical rigor.
Cell Type Deconvolution
Determining cell type-specific expression patterns from bulk tissue RNA-seq data.
Solution:
Developed a computational approach to map differentially expressed genes to cell type-specific expression databases, enabling identification of cell populations most affected in Alzheimer's.
Integration of Analysis Tools
Combining Python and R-based tools into a cohesive analysis pipeline.
Solution:
Created a seamless integration framework using rpy2, with careful management of data structures and memory to enable efficient data transfer between environments.
Interested in collaborating?
I'm always open to discussing new projects and opportunities.
Get in Touch