Phishing Detection using LSTM Architecture
Project Description
The Phishing Email Detection AI uses a Long Short-Term Memory (LSTM) neural network to identify and classify phishing emails with high accuracy. The system leverages natural language processing (NLP) techniques to analyze email content, focusing on sequential patterns and contextual relationships in text data. The model preprocesses raw email text by tokenizing and padding sequences to ensure uniform input for prediction. Trained on a labeled dataset of legitimate and phishing emails, this ML powered solution provides real-time predictions via an API built with Flask, making it deployable in production environments.
Use Cases
- Email Security Enhancement: Automatically detects and filters phishing emails within email platforms, reducing exposure to cyberattacks.
- Employee Training Programs: Provides datasets for phishing awareness training, enabling employees to identify and avoid malicious emails.
- Augmenting Enterprise Security Systems: Integrates with SIEM solutions to enhance email threat detection using NLP-based analysis.
- Automated Incident Response: Flags phishing attempts in real-time, enabling swift action by cybersecurity teams to mitigate risks.
- Seamless API Integration: Offers REST API endpoints for third-party tools like CRMs and helpdesk systems to validate emails in real-time.
ChatDB
Project Description
ChatDB is a dynamic database query generation platform designed to streamline interactions with SQL and MongoDB databases. It empowers developers, database administrators, and analysts to efficiently generate, test, and execute structured queries without manual scripting. The platform leverages pattern-based query templates to ensure robust, error-free operations, addressing diverse data retrieval, aggregation, and transformation needs.
Use Cases
- Data-Driven Development: Automates query generation for e-commerce and analytics platforms, enabling real-time user-specific reports using dynamic patterns.
- Database Optimization: Generates optimized queries for complex data operations, improving performance and reducing resource utilization.
- Cross-Platform Integration: Seamlessly works with both SQL and MongoDB databases, providing a unified interface for different database systems.
- Rapid Prototyping: Enables quick development and testing of database queries for new features and applications.
- Educational Tool: Serves as a learning platform for understanding database operations and query optimization techniques.
AiFRED
Project Overview
AiFRED serves as an all-in-one solution for classroom management and analytics, combining face recognition, hand-raise detection, and question relevance analysis to enhance the teaching and learning experience.
Core Features
- Attendance Management and Student Records:
- Uses face recognition (via OpenCV and face_recognition libraries) and embedding comparisons to identify students
- Stores detailed student information in MongoDB
- Allows instructors to create and manage courses
- Real-time Engagement and Hand-Raise Detection:
- Employs Mediapipe for hand tracking
- Logs hand raises in MongoDB for participation insights
- Question Relevance Analysis:
- Uses OpenAI APIs (GPT-4 and Whisper) for question classification
- Automatically transcribes and analyzes student questions
- Data Visualization and Analytics:
- Offers PyQt5 interface for metrics analysis
- Generates Matplotlib charts for performance insights
Use Cases
- Classroom Attendance and Monitoring:
- Automatic face recognition-based attendance
- Ideal for large classes requiring systematic tracking
- Participation Tracking:
- Logs hand raises and questions for engagement analysis
- Helps identify students needing intervention
- Data-Driven Teaching Adjustments:
- Real-time analytics on question relevance and engagement
- Identifies areas of confusion for content refinement
- Course Management:
- Supports multiple courses and department-wide rollout
- Cross-references trends across courses for curriculum optimization
Technical Features
- Face Recognition and Computer Vision:
- OpenCV and face_recognition for detection and matching
- Efficient embedding storage in MongoDB
- Hand-Raise Detection:
- Mediapipe Pose Estimation for landmark detection
- Threshold-based hand raise detection
- Audio Processing:
- PyAudio and wave for audio capture
- OpenAI Whisper for transcription
- Database Architecture:
- MongoDB with optimized indexes
- Transaction support for data integrity
Tech Stack
- Programming Language: Python 3.x
- UI & Frontend: PyQt5, Matplotlib
- Database: MongoDB with pymongo
- AI & ML: OpenAI GPT-4, Whisper, OpenCV, Mediapipe
- Additional Tools: face_recognition, NumPy, python-dotenv
iWonder
Project Description
iWonder is an iOS app designed to perform real-time sentiment analysis on Reddit discussions. It leverages advanced Natural Language Processing (NLP) and VADER sentiment analysis to assess user sentiments on topics such as product reviews, trends, and community opinions. Powered by a Python-based backend and a native iOS frontend, iWonder delivers actionable insights with a seamless user experience.
Use Cases
- Informed Consumer Decision-Making: Users analyze Reddit discussions to gauge community sentiment on products and services, with actionable recommendations helping users decide whether to buy, wait, or avoid specific items.
- Market Sentiment Analysis: Enables businesses to monitor Reddit discussions for insights into customer opinions on their offerings, identifying trends and highlighting critical feedback for strategic decision-making.
- Real-Time Sentiment Tracking: Tracks shifting sentiments in real time, making it ideal for monitoring reactions to product launches, updates, or announcements, offering timely insights to businesses for reactive strategies.
- Trend Monitoring for Market Research: Facilitates long-term sentiment tracking to identify patterns, seasonal trends, or changes in public perception, generating valuable datasets for research and predictive analytics.
Technical Features
- Sentiment Analysis Pipeline: Utilizes VADER Sentiment Analysis to classify Reddit discussions into positive, neutral, and negative sentiment categories.
- Reddit API Integration: Implements asyncpraw for asynchronous streaming and analysis of the latest Reddit posts and comments.
- Visual Data Representation: Features SwiftUI Charts to display sentiment distributions, offering users an intuitive understanding of trends.
- Backend Framework: Developed with FastAPI, enabling efficient processing and low-latency RESTful API communication between the backend and frontend.
- Mobile-Optimized Interface: Designed for iOS using SwiftUI, incorporating smooth animations and a user-friendly layout for a native app experience.
Enterprise Network Design
Project Description
Enterprise Network Design is a comprehensive solution engineered to support over 100 devices across 10+ subnets, focusing on scalability, reliability, security, and redundancy. The network is structured using a three-layer hierarchical topology—core, distribution, and access layers—to ensure efficient traffic management, robust fault tolerance, and optimized performance for enterprise-scale operations.
Technical Features
- Network Topology: Implements a hierarchical architecture with dedicated core, distribution, and access layers, facilitating scalability and streamlined traffic flow.
- Dynamic Routing Protocols: Employs OSPF (Open Shortest Path First) to ensure optimal path selection using Dijkstra's Algorithm, enhancing fault tolerance and minimizing downtime.
- VLAN Segmentation: Ensures secure and efficient data traffic segmentation across departments, safeguarding sensitive data and improving overall performance.
- Core Layer Design: Utilizes Cisco Catalyst 9300-24T switches for high-speed inter-VLAN routing and centralized traffic management.
- Distribution Layer: Features Cisco Catalyst 9200-24T switches with HSRP (Hot Standby Router Protocol) for seamless redundancy and fault tolerance during hardware failures.
- Access Layer: Deploys Cisco Catalyst 2960X-24PS-L switches for end-device connectivity, configured with VLANs and DHCP pools to streamline IP address management.
- Enterprise Servers: Includes PowerEdge R350 servers to centralize critical network services such as DNS, DHCP, database hosting, and file sharing.
- Security Architecture: Secures critical infrastructure using a Fortinet FortiGate firewall to enforce ACLs and provide intrusion prevention for DMZ-hosted services.
Use Cases
- Enterprise-Level Support: Designed to support over 100 PCs distributed across segmented VLANs, ensuring seamless communication and scalable infrastructure.
- Traffic Isolation: VLANs segment sensitive traffic to enhance security and prevent unauthorized access between departments.
- High Availability: HSRP ensures uninterrupted network connectivity by automatically transitioning the gateway role during failures.
- Dynamic Management: DHCP pools configured for each VLAN dynamically assign IP addresses, reducing administrative effort.
- Centralized Administration: Unified deployment of PowerEdge R350 servers simplifies maintenance and reduces operational costs.
- Enhanced Security: Fortinet firewalls enforce ACLs to restrict unauthorized access, safeguarding critical resources.
- Optimized Performance: OSPF provides efficient data routing by selecting the shortest path, minimizing latency.
- Cost-Efficient Design: Budget-conscious hardware selections deliver enterprise-grade performance at a reasonable cost.
DJARVIS
Project Overview
DJARVIS is an innovative gesture-controlled music player that transforms hand movements into music control commands, offering a hands-free, intuitive way to interact with Spotify and local media playback.
Core Features
- Gesture Recognition and Control:
- Utilizes Mediapipe Hands for precise hand landmark detection
- Supports multiple gesture types including swipes, palm positions, and hand raises
- Real-time gesture processing with optimized frame skipping
- Music Playback Control:
- Seamless Spotify Premium integration via Spotipy
- Alternative VLC support for local media playback
- Volume control through vertical hand movements
- Gesture Command Set:
- Open Palm: Play/Resume playback
- Closed Fist: Pause playback
- Horizontal Swipes: Next/Previous track navigation
- Vertical Movements: Volume adjustment
- Two Hands Raised: Application exit
Use Cases
- Music Performance and DJing:
- Hands-free music control for live performances
- Intuitive gesture-based track navigation
- Real-time volume adjustments
- Home Entertainment:
- Contactless music control from a distance
- Natural interaction with music playback
- Accessibility-friendly interface
Technical Features
- Tech Stack:
- Programming Language: Python 3.x
- Computer Vision: OpenCV, Mediapipe
- Music APIs: Spotipy (Spotify Web API)
- Media Playback: python-vlc
- Additional Tools: NumPy, python-dotenv