MD. ASIF HASAN

Developed a personalized news recommendation system leveraging Neural Collaborative Filtering (NCF) to combine content-based and collaborative filtering techniques. The system was designed to handle large-scale datasets and generate precise, real-time recommendations, enhancing user engagement through tailored news suggestions.

• Implemented Neural Collaborative Filtering (NCF): Developed an NCF model for user-specific recommendations by learning interactions between users and news items.
• Hybrid Recommendation Approach: Combined content-based and collaborative filtering to improve recommendation accuracy using both user and content data.
• Model Optimization: Optimized training with model checkpointing techniques for continuous training and automatic recovery across large datasets.
• Performance Evaluation: Evaluated the system using precision, recall, F1-score, and AUC to ensure it met user satisfaction and accuracy requirements.

Developed an AI-powered chess bot that utilizes reinforcement learning techniques such as Q-learning and Deep Q-Networks (DQN) to learn and improve its gameplay. The bot also incorporates Monte Carlo Tree Search (MCTS) to evaluate and select optimal moves. It continuously adapts to opponents, making it highly competitive on online chess platforms.

• Reinforcement Learning Implementation: Developed a chess bot using Q-learning and Deep Q-Networks (DQN) for strategy improvement.
• Monte Carlo Tree Search (MCTS): Implemented MCTS for advanced move evaluation, allowing the bot to plan ahead and make strategic decisions.
• Competitive Performance: The bot consistently outperforms less-experienced players in simulated and real online matches.
• User Interface: Developed an intuitive interface enabling players to challenge the bot at different difficulty levels.

ASICoin is a blockchain-based cryptocurrency project still under development. The project aims to integrate AI-driven fraud detection and post-quantum cryptographic security to ensure the future-proof nature of the network. Additionally, dApps for mobile banking, file transfers, and educational record management are in the planning stages.

• AI-Enhanced Fraud Detection: Developing machine learning-based systems for monitoring transactions and detecting anomalies.
• Post-Quantum Cryptography Research: Researching lattice-based encryption algorithms to prepare for quantum computing risks.
• Decentralized Applications (dApps): Implementing dApps for secure mobile banking, file transfers, and educational record management.
• Layer 2 Solutions: Exploring integration to improve scalability and transaction throughput for high-speed transactions.

This Inventory Management Software for Medicine Shop is designed to streamline the day-to-day operations of managing medicine stock, processing sales, and optimizing inventory levels. The system provides real-time data insights, automated purchase recommendations, and an intuitive user interface to facilitate efficient management of products and sales activities.

• Real-Time Stock Updates: Automatically updates stock and sales data.
• Sales Tracking: Generates reports on sales, profit margins, and top sellers.
• Automated Reorder Alerts: Notifies when stock levels fall below thresholds.
• Expiry Date Tracking: Alerts when products near expiration.

An efficient Sudoku solver leveraging backtracking, recursion, and constraint propagation to find solutions for 9x9 grids. Optimized for speed and accuracy, ensuring compliance with Sudoku rules while handling even the most challenging puzzles.

• High Accuracy: Solves 100% of valid Sudoku puzzles with systematic elimination.
• Optimized Speed: Uses constraint propagation and backtracking to reduce search space.
• Advanced Validation: Detects unsolvable puzzles before solving.
• Handles Extreme Cases: Solves puzzles with minimal clues, including the hardest ones.

A cutting-edge AI-driven trading bot designed for real-time market analysis and decision-making. This bot leverages deep learning techniques, such as LSTM (Long Short-Term Memory) networks, to predict future price movements based on historical data. Technical indicators like Ichimoku Cloud and Bollinger Bands are integrated to enhance the bot’s ability to perform technical analysis and identify trends. Additionally, Deep Q-Learning is implemented to allow the bot to adapt and refine its trading strategies autonomously by interacting with the market environment.

• LSTM for Time-Series Forecasting: Accurately predicts price movements.
• Ichimoku Cloud Analysis: Identifies market trends and support/resistance levels.
• Deep Q-Learning Integration: Enables intelligent decision-making for buy/hold/sell actions.
• Backtesting Strategies: Validates the bot’s performance across different market conditions.

Developed a sentiment analysis module to enhance cryptocurrency price prediction models by analyzing real-time news articles, social media posts, and public sentiment. Integrated sentiment scores with LSTM-based deep learning models to forecast price movements of Ethereum (ETH) based on historical Bitcoin (BTC) data.

• Sentiment Analysis on Cryptocurrency News: Applied NLP techniques like BERT and VADER to analyze crypto-related news.
• Data Pipeline Development: Built a pipeline to preprocess text data and extract sentiment scores and keywords.
• LSTM Model for Temporal Dependencies: Used LSTM models with sentiment-adjusted features for price prediction.
• Backtesting Strategies: Implemented backtesting to fine-tune model parameters and optimize performance.

This project uses Long Short-Term Memory (LSTM) networks to predict Ethereum (ETH) price trends by analyzing historical Bitcoin (BTC) data. It incorporates various technical indicators, including moving averages, Relative Strength Index (RSI), and Moving Average Convergence Divergence (MACD), to improve forecasting accuracy. Additionally, sentiment analysis is used to account for the impact of market news on price movements.

• LSTM-Based Deep Learning Model: Developed an LSTM model to predict Ethereum’s price based on Bitcoin data.
• Integration of Technical Indicators: Enhanced prediction accuracy with indicators like moving averages, RSI, and MACD.
• Model Optimization: Fine-tuned LSTM model using TensorFlow for better prediction performance.
• Backtesting System: Validated model performance using historical data for live trading simulations.

The Employee Management System is a comprehensive solution for managing employee data, payroll, attendance, and performance. It provides an easy-to-use interface for administrators to manage employee information, track attendance, and generate payroll reports, while ensuring data security and scalability.

• Employee Data Management: System for managing personal information, job roles, and contact details.
• Attendance Tracking: Automated system for clocking in/out and recording working hours.
• Payroll Generation: Automated payroll calculation based on attendance and performance metrics.
• Real-Time Reporting: Real-time reports on attendance, salary, and performance for HR managers.

The Lending Library Management System is a software solution for managing books, patrons, and lending transactions within a library. It provides features for checking out and returning books, tracking due dates, managing inventory, and generating reports on library activity.

• Book Inventory Management: System for managing book records including title, author, and availability.
• Patron Management: Module for managing patrons, tracking borrowed books, and updating personal info.
• Search & Filtering: Function for users to find books by title, author, genre, or availability.
• Due Date Tracking: Automatic tracking and notifications for overdue books and fines.

The Ride Sharing Simulation System aims to simulate and optimize the operation of a ride-sharing network. The system is designed to model real-world factors such as traffic conditions, vehicle availability, user demand, and route optimization to improve the efficiency and cost-effectiveness of ride-sharing services.

• Demand & Supply Modeling: Simulated ride requests and vehicle availability.
• Dynamic Pricing: Adjusted fares based on demand and traffic conditions.
• Vehicle Assignment Algorithm: Efficient vehicle-rider matching.
• Simulation Engine: Tested scenarios using Python and SimPy.

The Caesar Cipher Decoding Tool is a web-based application that attempts to decode a given sentence by applying various Caesar cipher shifts (from 0 to 25) and also includes a reverse string decoding method. It checks the decoded sentences for valid words from a predefined word list and highlights those words.

• Caesar Cipher Decoding: Tries to decode the sentence with all possible Caesar shifts and displays the results.
• Reverse String Decoding: Checks if reversing the input sentence produces any valid words.
• Word Validation: Highlights valid words in each decoded sentence using a predefined word list.
• Interactive Interface: Displays results dynamically and provides the number of valid words in each decoded sentence.

The Map Generation Tool allows users to plot points on an interactive coordinate system based on degree and step inputs. The tool calculates coordinates, generates points dynamically, and displays serial numbers, degrees, and steps for each point. It also features zooming, panning, and point info toggling.

• Interactive Map: Allows users to input degrees and steps to plot points on a coordinate system.
• Zoom & Pan: The map supports zooming and panning for navigation.
• Dynamic Point Generation: Points are plotted based on calculated coordinates, with each point labeled with serial numbers and input values.
• Multiple Points Input: Users can input multiple degree and step pairs for generating multiple points.
• Point Info Toggle: Users can toggle visibility of point information (degree and steps) for better clarity.