Led a team and spearheaded the creation of a data science platform for developing, deploying, maintaining, and monitoring time-series forecast models. This platform utilized technologies such as DataBricks, MLflow, Prefect2 for Dataflow, Pyspark, deltalake, and GitHub workflows for CI/CD pipelines.

• Led and mentored Junior data scientists, junior software developers and analysts to build data science models and write clean and scalable code.• Developed machine learning models for time-series analysis to forecast short-term energy demand.• Designed, developed, and implemented a machine learning framework that is currently used to produce multiple predictive models for well-log data. This framework allows the building of end-to-end ML pipelines by manipulating a set of configuration files. It performs dimensionality reduction, sampling strategy selection, multi-stage outliers’ detections, model development, model tuning, validation, and model performance evaluation. This framework produced three client-facing products and contributed to getting multi-million-dollar contracts. Maintained ML models’ lifecycles using Mlflow. • Designed and developed a rule-based framework for cleaning, transforming, merging, and summarizing well-log data. This framework allows product owners and analysts to rapidly develop tools based on subject matter expertise. • Deployed python dependency manager (poetry) and tuned it to meet the needs of frameworks described above.• Designed and developed Azure DevOps, GitHub Actions CI/CD pipelines for the framework described above.• Scaled and automated processes to run predictive models at scale using Databricks, Airflow and Perfect.Technologies: Python (OOP, TensorFlow, SKlearn, PyOD, LightGBM, XGBoost, Optuna, statsmodels, imblearn, scipy, lasio, NumPy, pandas, poetry, pipenv, multiprocessing, matplotlib, Plotly, Dash), Docker, Pycharm, visual code, Kubernetes, Apache Airflow, Prefect, DataBricks, Azure DevOps, git and GitHub.

• Designed and developed a Restful API to deploy well completion performance prediction tool• Developed well cost, well production curve and economics models to measure performance of completed well and deployed these modules as a micro-service.• Developed multiple ETL processes on ingested data from Azure datalake to be loaded into a GPU powered database that is used by frontend microservices.• Evaluated and profiled multiple algorithms to optimize our solution’s performance.Technologies: Python (NumPy, pandas, SKlearn, multiprocessing, flask, falcon, matplotlib, line-profile and memory-profiler), docker, git, Java, KINETICA.

• Designed and developed multiple algorithms and tools for processing and extracting features from geographical positioning systems and inertial navigation data.• Designed, developed, and implemented graph theory-based optimization algorithms and machine learning for multivariate predictive and clustering models to solve navigation related problems. Performed all steps of data analysis including features selection, dimensionality reduction, sampling strategy selection, model development, model tuning and validation (using k-fold cross-validation). • Designed and developed a multi-Threaded KPI tool to evaluate and assess the performance navigation solutions.Technologies: Python (NumPy, pandas, SKlearn, Keras, multiprocessing, matplotlib and math), Matlab, C/C++, UML 2.2, Git, SVN (version control system), Java.

Project: “Agent-based modelling as an innovative tool to simulate disease dynamics of wildlife-livestock interface to support intelligent decision-making."Tasks performed were:• Processed Real world spatial data including geophysical data and GPS tracking data using python scripting under ArcGIS.• Designed the multi-agent AI system. • Developed a cell based virtual environment to simulate the real-life landscape dynamics and used multi-agent system, and machine learning algorithms to model animal behaviour and disease transmission.• Visualized and analysed the results of the model.Technologies: UML 2.2, Git (version control system), Java, Python (NumPy, pandas), MATLAB, ArcGIS and NetLogo.

Project: “Modelling and simulating individuals’ mobility”: developed a decision support system that allows planners and policy makers to analyse and evaluate urban and transportation strategies.Tasks performed were: • Processed real world large data (millions of observations) and Performed Extract, Transform and Load (ETL) using python, ArcGis, Matlab and SQL server.• Developed and implemented models using data mining (filtering, sampling and cleansing, exploratory data analysis and model development using various machine learning models and results validation), Cellular Automata and multi-agent systems to extract knowledge and patterns about land use, individuals’ transport mode choice, to forecast land use change and transportation demands and to simulate traffic and assess its environmental impact. Developed and tuned uni and multi-label clustering and predictive machine learning models such as CART, MLP Neural Network, SVM, and KNN to successfully accomplish this project. • Visualized, mapped, and analysed the results of the developed models for QA/QC.Technologies: OOP, UML 2.2, incremental and iterative agile-like software development life cycle, Git, XML (XPath, XQuery and DTD), Java, JDBC, Swing, Weka (Java), Python (NumPy, pandas), SQL server (Stored procedure, Trigger and T-SQL), MATSIM and SUMO traffic simulators, MATLAB, Eclipse IDE, ArcGIS and GEOTOOLS.

Project: “Multi- level data Geo-visualization and dynamic modelling”: developed software for geocoding postal addresses information extracted from surveys.Tasks performed were: • Collected user requirements for engineering software that meets client/users’ needs.• Performed Extract, Transform and Load (ETL) in SQL Server.• Designed and developed a relational database for Luxembourg postal addresses.• Developed multi-threaded text data mining and record linkage algorithms.• Implemented graphical user interfaces and successfully deployed the software at two large public institutes in Luxembourg.Technologies: OOP, UML 2.0, Java, multi-threading in Java, JDBC, Swing, SQL server (Stored procedure, Trigger and T-SQL), XML (XPath, XQuery and XSD), Eclipse IDE, and Advanced installer for software packaging and deployment.

Project: "Development of a website for the Faculty of Science, Lebanese University using three-tier architecture."• Designed and developed SQL Server Relational database for the Lebanese University website. • Developed the Data access layer and Business logic layer of the website. Technologies: UML 2.0, SQL server, C#.Net, ADO.Net and Visual Studio

Insure the quality of service of "Map" and "team work" software by applying different software testing techniques, e.g. White box testing, black box testing.