Khaled Boukharouba, Phd Email & Phone Number

Paris, Fr

As the Data Science/Data Insights Director at Ogury, I lead three integral teams at the forefront of our strategic vision, focusing on personified targeting in a cookie-less and id-less environment. My role involves orchestrating the scaling of these initiatives, prioritizing tasks, and collaborating closely with the product teams to ensure seamless integration.A key aspect of my responsibilities is driving strategic initiatives and influencing decision-making across departments. I lead the research and development of machine learning, LLM and AI, integrating external data sources to enrich our capabilities further. By evangelizing and establishing standards for quality practices and processes, I instill a culture of excellence within my teams and across the organization.Navigating and influencing cross-functional teams comes naturally to me, as I proactively advocate best practices and lead by example. Ensuring compliance with data protection regulations and implementing robust data privacy policies are integral parts of my role. I foster teamwork, encouraging cooperation and leveraging complementary skills and expertise to drive success in our endeavors.

I led a dynamic team of data scientists and engineers focused on enhancing the delivery decision system. Our primary achievement was the development and implementation of a groundbreaking delivery algorithm, tailored to optimize volume while striking a balance between the interests of Ogury, publishers, and brands.One of our key innovations was the establishment of a comprehensive simulation and A/B test environment, enabling us to rigorously test and refine algorithms for optimal performance. Central to our approach was the creation of a suite of targeting models designed to operate without relying on user data, thus aligning with evolving privacy standards and regulations.In addition to spearheading algorithmic advancements, I formulated and managed a range of data-driven projects aimed at driving business growth and efficacy. This involved orchestrating the collection and cleansing of data from diverse sources, facilitating its utilization by junior data professionals across various domains including web crawling, reviews, and more.Furthermore, I played a pivotal role in scaling semantic data, expanding the scope of survey data, and harnessing the power of AI to uncover actionable insights across different languages and countries.

Paris 08, Île-De-France, France

I was in charge of a team of data scientists and engineers dedicated to data augmentation. Our primary focus was on enhancing signals gathered from mobile devices, utilizing advanced techniques to extract valuable insights and strengthen our targeting capabilities.Navigating the realm of big data, we tackled complex challenges with a blend of machine learning techniques and innovative approaches. I led the design and execution of large-scale data analysis projects, guiding my team in developing novel methodologies to address intricate analytical hurdles.A key aspect of my role was effectively communicating these complex ideas to both technical and non-technical stakeholders, ensuring alignment and understanding across diverse teams and departments. By fostering a culture of collaboration and innovation, we drove impactful advancements that contributed to the success of our organization.

Région De Paris, France

In 2016, I was promoted to the position of chief data scientist CDS/CTO @ Trinov. I am in charge of leading and supervising the team of data scientists, as well as a PhD student (CIFRE). I have successfully completed many of the tasks entrusted to me such as: - Drive the data strategy of TRINOV and build data factory;- Participate in the definition of the roadmap, brainstorming and pre-sales; - Conduct many “Proof of Concept” (POC) projects;- Participate in the drafting of commercial proposals and pricing; - Regularly carry out technological watch on new developments in "data science" and "open data";- Manage and animate the Kanban method.

Région De Paris, France

In 2013, I joined TRINOV, an innovative startup specialized in waste management. At Trinov, we develop different solutions (software) for small, medium and large size companies such as Carrefour, Auchan, Leader Price, Vinci, etc., to optimize their waste management and help them reduce the cost. When I first joined Trinov, I was the only data scientist there, hence, I had the chance to create a data team and build a data culture from scratch. I developed, during my first years, new optimization, modeling and prediction techniques as well as visualization techniques.

Région De Paris, France

I joined the SIGMA Laboratory (SIGnaux, Templates, Statistical Learning) of CNRS as a research engineer. During my time there, I developed algorithms and modules for flash floods prediction using databases provided by our partners including "Météo France". Flash floods are floods that can occur in a short period of time (very few hours). We proposed a new machine learning approach to flash flood forecasting in the absence of rainfall forecasts, based on the agglomerative hierarchical clustering of flood events. Each cluster contains events which models have similar behavior. Specific Support Vector Regression (SVR) models were trained using data from each cluster. Different techniques such as LSSVM and PWA were also tested.

- Lectures, practical works, labs about information systems and computer engineering for bachelor and master’s students- Correction and supervision of examinations

Douai

I worked as a research engineer in the "Computer and Automation" department of “Mines de Douai” where I developed new algorithms and modules for consumers behavior modeling based on their web purchases. The purpose was to identify potential consumers based on the time spent browsing the website, the amount of the shopping basket, the number of connections, etc.

Douai

I worked on the identification of a special class of hybrid systems, namely, the class of PieceWise Affine (PWA) systems from input-output data. The identification of PWA models is a challenging problem. It involves the estimation of both the parameters of the affine sub-models, and the coefficients of the hyperplanes defining the partition of the state + input set. I proposed new methods for identifying PWA models from data. The solution includes the estimation of the number of sub-models, the identification of the parameter vectors that describe the set of sub-models and the determination of the bounding hyperplanes of the polyhedral regions associated to the sub-models. I also proposed a recursive algorithm for identifying PieceWise Affine systems (PWA) and PieceWise nonlinear systems where the parameters of the sub-models and the regions can vary over time. A recursive LS-SVM technique is proposed for the recursive update of the parameters of each sub-model. The adaptation of the parameters of the regions is ensured by an online multi-category support vector classifier. I wrote several scientific publications in conferences and international journals during my PhD in machine learning.

Région De Lille, France

- Undertaking hundreds of hours of supervised and practical works (travaux dirigés TD et travaux pratiques TP) in different fields of study (computer science and Computer Engineering) for undergraduate and master’s students. I also supervised many students in their master’s project. - Correction and supervision of examinations related to the fields of study.

Douai

I was in charge of providing supervised and practical works (travaux dirigés TD et travaux pratiques TP) in different fields of study (computer science and Computer Engineering) for student-engineers.

Douai

I developed an online recursive clustering algorithm based on incremental and decremental Support Vector Machine (SVM). The online clustering task consists in updating the data partition over time, according to some dynamic phenomena such as: drifting, creation, fusion or elimination. The proposed algorithm can achieve an efficient clustering and is able to give exact results compared to QP approach. The first-order KKT conditions of each cluster model (each model is a “one-class SVM”) are kept satisfied when adding/eliminating a new/old data.

Doctor Of Philosophy (Ph.D.), Machine Learning

Master 2 (M2), Automatique, Génie Informatique Et Images, Mention Bien (Major De Promotion),