🎓 Computer Science — Artificial Intelligence Track at An-Najah National University

🤖 Focused on building practical AI systems across Machine Learning, Deep Learning, Natural Language Processing, Computer Vision, Large Language Models, and Data Mining.

🚀 I enjoy taking AI projects from data preparation and model development to evaluation, optimization, and interactive applications.

🏆 1st Place — Build With AI Datathon

• 🎓 Computer Science student specializing in Artificial Intelligence
• 🧠 Interested in Machine Learning, Deep Learning, NLP, LLMs, and Computer Vision
• 📊 Experienced with classification, prediction, clustering, recommendation systems, and data mining
• 👁️ Built computer vision systems using YOLOv8, ByteTrack, OpenCV, and PyTorch
• 💬 Worked with Transformers, Mistral-7B, LoRA fine-tuning, TF-IDF, SVM, and sequence labeling
• 📈 Built end-to-end data science and business intelligence applications using Scikit-learn, XGBoost, Pandas, and Streamlit
• 🤝 Experienced in collaborative academic projects and team-based AI development
• 🔎 Continuously improving my skills through practical projects, research, and advanced AI coursework

An NLP and Large Language Model project for automatically evaluating prompt quality.

• Fine-tuned Mistral-7B using LoRA
• Built with Unsloth and Transformer-based workflows
• Evaluates prompts using structured quality criteria
• Uses metrics including Accuracy, MAE, RMSE, Spearman correlation, and BERTScore
• Includes an interactive Streamlit application

Technologies:
Python Mistral-7B LoRA Transformers Unsloth NLP Streamlit

A computer vision system for real-time vehicle detection, classification, tracking, and speed estimation.

• Vehicle detection using YOLOv8
• Multi-object tracking using ByteTrack
• Vehicle classification for cars, trucks, and buses
• Continuous vehicle speed estimation
• Fine-tuning using traffic data
• Video processing and visualization with OpenCV

Technologies:
Python YOLOv8 ByteTrack OpenCV PyTorch Computer Vision

An interactive retail analytics and business intelligence platform combining multiple data mining techniques.

Main components include:

• Customer Segmentation
• RFM Analysis
• Market Basket Analysis
• Apriori Association Rules
• Sales Trend Analysis
• Anomaly Detection
• Recommendation System
• Customer Profiling
• Interactive Business Reports

Technologies:
Python Pandas Scikit-learn Apriori Data Mining Streamlit

A Java-based NLP project for Arabic Named Entity Recognition using a Maximum Entropy Markov Model.

• Sequence labeling with MEMM
• Viterbi decoding
• Contextual and lexical feature engineering
• Morphological features for Arabic text
• Named entity identification with emphasis on person names

Technologies:
Java NLP MEMM Viterbi Sequence Labeling Feature Engineering

A traditional machine learning NLP pipeline for classifying movie reviews as positive or negative.

• Text preprocessing and stemming
• TF-IDF using unigrams and bigrams
• Multinomial Naive Bayes
• Linear SVM
• Random Forest
• Cross-validation and GridSearchCV
• Final Linear SVM test accuracy of approximately 87%

Technologies:
Python NLTK TF-IDF SVM Random Forest Scikit-learn

A structured collection of Advanced Machine Learning and Deep Learning coursework.

Implementation and training of a logistic regression model using PyTorch.

Deep learning-based character recognition using the EMNIST dataset.

Image classification using pretrained convolutional neural networks:

• ResNet50
• MobileNet

with transfer learning and model comparison.

Topics:
PyTorch Deep Learning EMNIST Computer Vision CNN ResNet50 MobileNet Transfer Learning

A collection of practical Data Mining coursework covering multiple unsupervised learning and pattern discovery techniques.

• K-Means
• Agglomerative Clustering
• DBSCAN
• Silhouette Score
• Cluster evaluation using the Iris dataset

• Association Rule Mining
• Frequent pattern analysis
• Principal Component Analysis
• Covariance Matrix approach
• Singular Value Decomposition

Topics:
Data Mining Clustering K-Means DBSCAN Association Rules PCA SVD

• Machine Learning
• Deep Learning
• Natural Language Processing
• Large Language Models
• Computer Vision
• Data Mining
• Recommendation Systems
• Predictive Analytics
• Artificial Intelligence Applications

• 🥉 1st Place — Build With AI Datathon
• 🎓 AI Programming with Python & TensorFlow Nanodegree — Udacity
• 🤖 Built multiple end-to-end Machine Learning, NLP, Computer Vision, and Data Mining applications
• 🧠 Practical experience with traditional ML, deep learning, LLM fine-tuning, and transfer learning
• 👥 Contributed to and led collaborative university AI projects

I am particularly interested in:

• Large Language Models
• LLM Evaluation
• Prompt Quality Evaluation
• Multi-Agent AI Systems
• NLP Evaluation Metrics
• Computer Vision
• Applied Machine Learning

My recent work includes experimentation with Mistral-7B, LoRA fine-tuning, LLM-based evaluation, and multi-agent social simulation research.

📧 Email: karshafie@gmail.com

📍 Palestine

I am interested in opportunities related to:

AI Engineering • Machine Learning • NLP • Computer Vision • Deep Learning • Data Science

I am also open to collaboration on practical AI projects, research, and intelligent software applications.

⭐ Thanks for visiting my profile!