Sana Elwaar

Led the strategy and development of a no-code chatbot orchestration platform from POC to production (Product market fit for one platform serving 100+ use cases for a French Banking Group) • AI (and GenAI) specific risks handling and mitigation (moderation of both input user question and LLM output, non-regression tests, hallucination tracking, correctness measuring and red-teamming). • Design and delivery of the GenAI evaluation platform (validation data life cycle management, testing acceleration and alerting automation) • Worked closely with data scientists, designers, use case product owners and users in order to develop user-friendly features without giving up the AI performance. Discovery, training and validation of the AI components of the first Use Case of the platform (NLP, search engine and retrieval, NLG and moderation, NLU and intention detection) • Reduction of the time to market to under 3 weeks (divided by 4). • Launch of a B2C chatbot for client assistance for 5 million users

- AI program mentor: co-defined, developed and deployed an AI assistant for telecom dispatching and a cost-effective computer vision system for worksite safety. Mentored internal resources towards data science proficiency. - Technical due diligence: technical assessment and investigation in order to give key elements for the strategic decision-making. - Implemented machine learning algorithms for contamination pattern identification and threat prediction in cybersecurity logs. Promoted data-driven approaches and agile methodologies (Shape up) within the team. - Contributed to the internal structuring efforts: responsible of the communication on the web site, guidelines, content strategy definition with stakeholders, content creation pipeline initiation and follow up, edition etc.

Full immersion in the international environment, Focus on data science for CRM (Customer Relationship Management) • Support of our client in the transformation of his CRM strategy: analysis of the business evolution and the value of the customers in Japan, Customer Segmentation and strategic recommendations adapted to the cultural specificities and the strategic objectives. (Regression trees, market basket analysis, products recommendation) • Development of a tool to automate basic CRM analysis, 4 times faster, easily reusable thanks to the code in modules. (VBA, R, Data visualization in Excel) • Development of an NLP tool to valorise customer related data in the Chinese social media platforms: data Webscrapping, preprocessing, topic detection and opinion-mining of Mandarin comments. The tool will then mainly perform a Data Vectorization using Word2vec Embedding and then a sentiment classification (gradient boosting model). The sentiment classification accuracy exceeded the 90% in the cross validation test (based on 2000 comments). • Results sharing during knowledge management sessions, code tutorials delivering for further uses of the 2 R&D tools developed during this internship.

Collection of clients' comments and analysis of the opinions in order to determine the reputation of a given brand. • Webscaping • Textmining- opinionmining (in french) • Data Visualization

Operations of Extract, Transform, Load and Visualization • Treated variety of data inputs, applied unsupervised data mining algorithms to cluster the clients of CIB BNP Paribas and quantify their performances. • Produced steering dashboards for the top management, and automate data quality reports creation.

Internship with Safran, an international high-technology leader in aerospace, defense and security. • Innovative international project aiming to design the cabin of the aircraft of the future. • Market and Feasibility studies, • Preliminary 3D Computer Aided Designs and product compatibility studies of the new innovative products. The value proposition was to supply power and data in the passengers' cabin in a reconfigurable way. • Internship held in the laboratories of the "Escola Politecnica da Universidade de São Paulo" (Poli-USP).

Internship with Tunisie Télécom, is the incumbent telecom operator in Tunisia. - Worked in a team of 7 technicians in a local telephony distributer. - Maintained the telecommunication hardware in the cell according to the instructions of the interactive dialogue interface.