Dr. Ehab HASSAN

Generative AI-based job description generation: Build an generative AI chatbot that automatically generates job descriptions in several natural languages based on keywords given in the user query. - Build generative AI systems for job description generation utilizing OpenAI’s models. - Preparation and vectorization of historical data using LangChain and embedding-ada-002. - Create a vector store with OpenAIEmbedding and FAIS in order to build a powerful information retrieval system (RAG). - Prompt Engineering: develop and optimize prompts in order to effectively query our generative AI models and get the desired response. - Build an agent to identify the question language and answer in the same language. - Build a traduction agent to ensure the traduction step. - Tools: Python, OpenAI, Embedding-ada-2, Faiss, RAG, LLM, Streamlit.

- Understand clients needs and propose methodologies that effectively meet their needs. - Help clients to understand data and identify innovative and varied use cases. - Support clients throughout the Data Science project cycle, from data collection to the production of machine learning models. - Manage the relationship with clients. - Use cases: 1. Ticketing System: Build an AI system that automatically reads an incident ticket written in spanish and suggests one or more solutions for this incident. -Tools: Python, OpenAI, Langchain, LLM, RAG, FastAPI 2. Contractual Risk Analysis: Build an algorithm of artificial intelligence based on Natural Language Processing that helps project managers to spot the dealbreakers, fill the risk analysis file and assist them in contract management. - Tools: Python, Spacy, Gensim, Word2Vec, Transformers, LLM, Streamlit. 3. Chatbot: Building a chatbot with reading function for open tickets written in French. The chatbot firstly read the incident ticket, ask questions in order to better understanding the issue described into the ticket, then executing an AI System in order to suggests one or more solutions for the ticket. - Tools: Python, Spacy, Gensim, Word2Vec, Transformers, Watsonx 4. Automated system for tenders pricing: Build an AI system for costing calls for tenders. - Tools: Python, Spacy, Gensim, Word2Vec, Transformers 5. Energy Consumption Optimization: Use AI methods to optimize the energy consumption of a smart building equipped with a digital twins. Tools: Python, DataRobot. 6. Accidentology: Use transport data to predict and understand accidents. - Tools: Python, DataRobot. 7. Predictive Maintenance: Use data from a refinery to indicate the reason for a motor current alert. - Tools: Python, DataRobot, Paxata.

1. Use Deep Learning for images classification and objects detection: - Use Machine Learning methods to build image classification models. - Implement CNNs method (Convolutional Neural Networks) to build image classification models. - Use Faster-RCNN method to build a model for detecting objets on images. - Perform and test these tasks on DataIku. - Tools: Python(Scikit-learn, TensorFlow, Keras), DataIku. 2. Sentiment Analysis for French Texts: - Apply NLP methods (Tokenization, Lemmatization, POS, N-Gram, Stop-Words,...) to identify and extract text features. - Use Machine Learning methods (SVM, NB, Random Forest, Neural Networks, ...) to build sentiments classification models. - Perform and test these tasks on DataIku. - Tools: Python(Scikit-learn, Spacy, NLTK), DataIku. 3. Sentiment Analysis for English Texts: - Apply NLP methods (Tokenization, Lemmatization, POS, N-Gram, Stop-Words,...) to identify and extract text features. - Create and train a Word2Vec model to identify text features. - Use Machine Learning methods (SVM, NB, Random Forest, Neural Networks, ...) to build sentiments classification models. - Implement CNNs method (Convolutional Neural Networks) to build sentiment classification models based on Deep Learning. - Perform and test these tasks on DataIku. - Tools: Python(Scikit-learn, Spacy, NLTK, Keras, Gensim), PySpark, DataIku

Program "Machine Driven Big Data", Altran Research 1- Manager of the project "GOTTRA++": GOTTRA++ is a research project that aims to design and develop a set of tool that uses NLP, text mining, and machine learning technologies in order to automate all the steps of TRA test projects in Agile environment. - Contribution to the development and design of the tool GOTTRA++. - Use NLP and text mining methods to extract use cases and test scenarios from functional specifications. - Apply machine learning algorithm to identify automatizable scenarios - Staffing and integrating consultants on the GOTTRA++ project. - Assignment of tasks for the integrated consultants. - Follow the work of the consultants and the project progress. - Supervise trainee students on the project. - Validate the results of consultants and trainee works. - Java/JEE, Python (NLTK, scikit-learn), SQL 2- Manager of the project "GOTTRA": GOTTRA is a research project that aims to design and develop a system and a tool in order to help testers to find the good estimation of charge. GOTTRA in his actual version uses machine learning technologies in order to estimate and predict the charge of test projects based on old test projects. - Contribution to the development and design of the tool GOTTRA. - Implementation of machine learning algorithm - Staffing and integrating consultants on the GOTTRA project. - Assignment of tasks for the integrated consultants. - Follow the work of the consultants and the project progress. - Supervise trainee students on the project. - Validate the results of consultants and trainee works. - Java/JEE, Weka, SQL

PhD Subject: Event-Based Recognition Of Lived Experiences In User Reviews. Realizations : - Machine Reading: Proposal of a method for transforming user reviews into their semantic representations (RDF / OWL) in order to extract knowledge (named entities, senses, taxonomies, relations, events) form them. - Events Extraction: Development of an event extraction method from user reviews using NLP and semantic Web techniques. - Sentiment Analysis: Proposition and development of a classification approach allowing to study the correlation between user opinions about services/products and the ranking attributed by them. - Lived Experiences Identification: Proposal and development of a system allowing to identify user reviews containing lived experiences using text mining, semantic Web, and NLP methods. - Lived Experiences Extraction: Proposal, conception, and development of a RESTful application allowing to extract lived experiences from user reviews in the tourist field, Implemented using Java and J2EE. - Spam Detection: Development of a classification approach for detecting spam user reviews based on machine learning, NLP, and semantic Web methods. - Text Summarization: Proposition and development of an algorithm for summarizing several user reviews based on machine learning, semantic Web, and NLP methods. - JAVA, JAVA EE, RDF/OWL, SPARQL, FRED

Teaching Courses: - Knowledge Representation (TD/TP) - Semantic Web (TD/TP) - Data Processing (TD/TP) - Data Bases (TD/TP) - Logic (TD/TP) - Systems and Networks (TD/TP) - IT Administration (TD/TP) - Language C (TD/TP) - Computer Algorithms (TD/TP) - Imperative Programming (TD/TP)

Subject : Semantic Enrichment of documents with annotations for information retrieval task. Realizations : - Development of an annotation tool for legal documents. - Semantic enrichment of documents with annotations in order to facilitate information retrieval and navigation in the documentary collection. - Tool: Java, RDF/OWL, SKOS, TreeTagger.

Subject: Measuring the effectiveness of the business rules of Google referencing using Data Mining methods. Realizations : - Analyze Google "Adwords" data to find and extract the most frequent set of keywords. - Analyze the Data provided by the company "préférencement" to understand users behaviors. - Match Google "AdWords" data with "préférencement" Database in order to detect the most effective keywords. - Use data mining methods (classification and clustering) in order to indicate keywords to be purchased and obtain a better referencing in Google. - Java, SQL, Weka platform, Tanagra Platform, Google Adwords.

Subject: Interactive Display of Critical Edition (Sanskrit case) Realizations : - Proposal and development of an approach to automatically pass from the critical edition results to its XML version and then to its interactive version. - Java, XML, DOM, XSLT, HTML, CSS

PhD Subject: Event-Based Recognition Of Lived Experiences In User Reviews. Publications : 5 research papers, 4 international conferences (EKAW, ISWC, SEMANTiCS, ICONIP), and 1 national conference Keywords : Machine Reading, NLP, Machine Learning, Knowledge Engineering, Semantic Web

Data Mining/Machine Learning, Semantic Web, Data Warehouse, Advanced Data Bases