I'm AI Engineer Menna Mostafa

with a strong background in Computer Science and Artificial Intelligence, specializing in computer vision, NLP, and LLMs. Experienced in developing innovative AI solutions at Uqudo and Devisionx, with expertise in Python, TensorFlow, and PyTorch. Committed to creating impactful AI technologies that drive positive change.

Get started Learn more

Experience

AI Engineer at DevisionX

implemented image and video analytics algorithms, evaluating performance on large datasets. Integrated the machine learning platform with ClearML and MLflow. Created and reviewed documentation and software development plans.

AI Engineer at uqudo

achieved a competitive edge with advanced document support for KYC through
enhanced text detection. Improved model accuracy with synthetic data preparation, achieving 95% accuracy on real-world validation datasets. Ensured optimal performance with the deployment of text detection models in ONNX format.

AI Intern at Fixed Solutions

Developed Arabic OCR and GenAI chatbot applications, integrating Gemini for creative text generation. Fine-tuned a Git model using Arabic caption data to enhance performance and accuracy.

AI Intern at Samsung

Applied machine learning and deep learning techniques using TensorFlow and PyTorch in various projects, including Visual QuestionAnswering and Face Detection.

Skills

Programming Languages:

Python, C++, SQL, R, JavaScript

Deep Learning Frameworks

TensorFlow, PyTorch, Keras

Libraries & Tools

NumPy, Pandas, Scikit-learn, OpenCV, Ultralytics , Pillow, Matplotlib, Seaborn, NLTK,
Git, Docker, Tableau, PowerBI,
Gradio, Flusk Api, Streamlit

Projects

Smart Vision App

Developed an Egyptian Cash Reader for banknote recognition, manually collected and annotated the dataset, integrated objectdetection with OCR and QR code.

Generative Video Captioning

Implemented scene detection, applied Llama2 for title generation, created a caption dataset to fine-tune the Git model, integrated models into a Streamlit demo

Comparative Analysis of AI Chatbots for Document Retrieval

Analyzed PaLM, GPT-2, and FLAN-T5 for document context retrieval and accuracy evaluation, provided an analysis report, integratedmodels into a Gradio demo.

Padel Smart System

Developed an AI-driven system to enhance training environments by implementing stroke type detection, court and player detection, Position estimation, pose estimation and ball tracking.

Auto-OCR Vision

Fine-tuned YOLOv8 and Phi-3 models for Arabic text detection and extraction, integrated with Gradio for deployment