2022 - 2023: Scalable Text and Analytics Medical Insight Clustering Engine with LLMs & Generative AI

Role: Expert Data Scientist
Customer: Merck Group KGaA,

Tasks:

• Led the development of Merck?s Medical Clustering Insights Engine that achieved recognition in top 3 innovative AI Solutions across the group.
• Solved the difficulty in handling large volumes of text data from various data sources for efficient analysis,and in achieving high accuracy when investigating critical therapeutic areas as Oncology and Immunology due to unstructured clinical data and overcoming the Limited capability of traditional NLP techniques for the needed complex tasks .
• Fine-tuned BERT-based LLMs: Pre-trained Large Language Models (LLMs) BERT-based are further customized and fine-tuned on the Pharmaceutical data (Hugging Face Transformers, PyTorch)
• Developed Clustering for TA Analysis: The engine utilized clustering techniques into multi layer system to group similar clinical findings using( scikit-learn, TensorFlow, PyTorch), allowing Business units managers and other stakeholders to focus their investigations on the relevant data points.

2021 - 2021: Fraudulent Patent Detection in different languages with fine-tuned LLMs

Role: Senior Data Scientist
Customer: NLP , Predori GmbH,

Tasks:

• Pioneered state-of-the-art (SOTA) modules with fine-tuned Bert based LLMs, capable of accurately identifying fraudulent patents together with machine translation algorithm that Contributed in receiving an award in the Start-up Baden-Württemberg competition.
• The challenge in traditional approaches across various domains: identifying fraudulent patents faces difficulties with complex languages and obfuscated details, manual processes impede the efficiency of patent verification, and traditional machine translation proves resource-intensive and costly for handling large volumes of text.
• Information extraction techniques to identify key details from patent text (e.g., inventor names, dates, claims).
• Containerization with Docker and Deployment using AWS Lambda functions through Packaging the fine-tuned machine translation modules and its dependencies into container allows efficient deployment management.

2020 - 2020: Streamlined Data Management and Analysis over GCP

Role: Senior Data Scientist
Customer: Indeed

Tasks:

• Designed and implemented a scalable data scraping solution using (Selenium and BeautifulSoup), reducing data acquisition time by 40% for the B2B Data Products team.
• everaged GCP for scalable data management through Establishing a (BigQuery) data warehouse for storing and accessing data and Integrating pipelines with GCP (Dataflow) for automated data ingestion.
• This project tackles the challenges of ensuring scrapped data quality, automating data workflows, and facilitating collaborative data access in one comprehensive solution.
• Improved data quality through automated validation and cleaning (pyspark and python)
• Streamlined data ingestion, transformation, and loading to BigQuery via automated Dataflow pipelines.
• Enhanced collaboration through centralized and accessible data storage on GCP Buckets.

2019 - 2019: Squad Lead in Vodafone?s Digital Transformation Team

Role: Senior Data Scientist
Customer: Vodafone

Tasks:

• Squad Lead in Vodafone?s Digital Transformation Team responsible for different Data science and Analytics use-cases implementation and their related migration into GCP,in different local market structure for Customer Value Management, Network Performance Enhancement, and Customer Experience Improvement.
• Using Python/PySpark and Scala for data manipulation and analysis, modelling on Google Cloud Platform. This resulted in a 15% increase in customer satisfaction and a 20% reduction in network downtime.
• Received Vodafone Knowledge Sharing Award in recognition of effectively mentoring and transferring technical knowledge to the team, contributing to their professional development.

2018 - 2018: Developed and delivered comprehensive supplementary materials

Role: Teaching Assistant - part time
Customer: Alexandria University

Tasks:

• Developed and delivered comprehensive supplementary materials, including lab manuals, tutorials, and cheat sheets, tailored to the specific content of the project for undergraduate Data Science laboratory sessions.
• Explained unsupervised learning concepts like clustering and dimensionality reduction, providing basic examples and highlighting their applications, and explaining mathematics behind them.
• PGuided students through practical exercises with implementing and evaluating these algorithms in Python libraries like scikit-learn. and Neural Networks with Keras.

2017 - 2018: Led the Data Science team at the Andalusia Group, driving business value through data-driven solutions

Role: Senior Data Scientist
Customer: Andalusia Group for Medical Services

Tasks:

• Led the Data Science team at the Andalusia Group, driving business value through data-driven solutions.
• Analyzed Electronic Medical Records (EMRs) to improve physician efficiency in Health Risk Assessment
• Developed prediction model to identify patients at high risk of requiring specific medical interventions (operations, services, or drugs).
• Enhanced patient segmentation and targeting by Performing patient segmentation analysis using various methods such as loyalty analysis, demographic clustering, and medical status analysis. by clustering algorithms to segment patients based on specific characteristics and behavior

2016 - 2016: Designed and implemented a comprehensive AI solution for Sporting Clubs MS

Role: Data Scientist
Customer: ProGaze

Tasks:

• Designed and implemented a comprehensive AI solution for Sporting Clubs MS, seamlessly integrating with their existing Microsoft Azure machine learning models to generate business predictions in areas like player performance, team success, and fan engagement.
• Settled the architectural design of a comprehensive AI solution tailored for Odoo ERP, by Developing machine learning modules that leveraged existing Odoo data models (MVC) to generate meaningful insights and improve operational efficiency.

2014 - 2015: Increased accuracy with Lasso regression and RMSE metric ensured reliable predictions

Role: Junior Data Scientist
Customer: Bojakt

Tasks:

• Developed house pricing and land evaluation engine with AI-driven predictions and recommendations, to empower clients to make informed investment decisions based on future market insights.