Final_Resume_AGRAT_Mohammed_compressed.pdf

AGRAT MOHAMMED Junior Data Scientist | Python Developer +212-6570664666 [email protected] https://www.linkedin.com/in/mohammed-agrat Morocco , Casablanca...

AGRAT MOHAMMED Junior Data Scientist | Python Developer +212-6570664666 [email protected] https://www.linkedin.com/in/mohammed-agrat Morocco , Casablanca Summary Junior Data Scientist with expertise in machine learning, predictive modeling, and statistical analysis, seeking to contribute to data-driven decision-making processes. Proficient in Python and familiar with ML libraries such as pandas, scikit-learn, and TensorFlow. Passionate about integrating data solutions into workflows, enhancing business efficiency through predictive analytics, ensuring high data quality, and comprehensive documentation. Strong analytical and problem-solving skills with the ability to work effectively in a team. PROFESSIONAL EXPERIENCE Capgemini Engineering Morocco,Casablanca Data Scientist Intern 03/2024 - 07/2024 Project : Development of a Classification Model for Temporary Signal Obstacles Implemented data preprocessing techniques to clean and transform raw data, ensuring data quality for analysis and predictive modeling. Developed and implemented predictive models using advanced classification algorithms to accurately classify obstacles on road surfaces, resulting in improved obstacle detection and enhanced traffic safety. Optimized predictive models and conducted comprehensive evaluations using various performance metrics, enhancing model accuracy and robustness in diverse road conditions. Deployed models in Glyphwork NCode software to streamline model integration and operations. Documented critical parts of the code and global workflows , Collaborated with the Signal Processing team to ensure model alignment with project objectives. Technocolabs Softwares India , Remotly Machine Learning Intern 07/2023 - 09/2023 Project: Mortgage Prepayment Risk Analysis Developed a predictive model using XGBoost Classifier to analyze mortgage prepayment risk, achieving an accuracy rate of 87%. Implemented a user-friendly web application on AWS EC2 for risk assessment, ensuring high availability and accessibility for end-users.. Utilized Docker to containerize the application, significantly enhancing scalability and reliability. Collaborated with an international team to ensure project success, facilitating cross-cultural knowledge sharing and effective teamwork. Education ENSAM Casablanca Casablanca Engineering Degree in Artificial Intelligence and Computer Science 2021 - 2024 CPGE Meknes Preparatory Classes for Engineering Schools (CNC) in Mathematics and Physics (MP) 2018 - 2021 Skills Programming Languages: Proficiency in Python Machine learning | Deep Learning| NLP:: scikit-learn TensorFlow pandas numpy OpenCV Pytorch Transformers LLM Flask Data Science & Modeling: Predictive Modeling Statistical Analysis Classification Algorithms Data Preprocessing Model Evaluation Metrics Monitoring Models Database Management:: Relational and Non-relational Databases Data Visualization: Power BI Matplotlib Plotly Knowledge of Cloud Platforms: AWS CI/CD Git/Github Docker Soft Skills: Problem-solving Critical Thinking Teamwork Creativity Communication  Projects Image Sentiment Classification Collected images data through web scraping and organized into 5 classes & developed a deep image classifier using Keras and TensorFlow, achieving an accuracy improvement from 52% to 82% using transfer learning with ResNet50. Resume Application Tracking System(ATS) Using LLM Model Developed ATS-compatible resumes using the LIM Model with Gemin Pro, optimizing formatting and keywords to improve ATS parsing accuracy US Visa Prediction : MLOPS Production Machine Learning end to end Designed and implemented robust pipelines for data processing, model training, and evaluation, ensuring a seamless end-to-end workflow Employed Evendently AI for continuous monitoring and performance tracking of deployed models, ensuring high accuracy and timely updates. Product Recommendations: Visually Similar Content Filtering using KNNs. Developed a content-based recommendation system to suggest products with visually similar features using K-Nearest Neighbors algorithms. This project demonstrated content filtering by leveraging item descriptions, even without user profile data, akin to recommendations seen on shopping websites. Image Super Resolution using Autoencoders Enhanced image resolution from low to high using autoencoders, applied to surveillance for facial recognition and identification on low-resolution images. Implemented super resolution image reconstruction to recover high-resolution images from degraded low-resolution inputs, tackling blur and noise issues.