Dylan Sadoudi

Drive the conception deployment of production level Retrieval-Augmented Generation (RAG) solutions in various industries, focusing on impactful and efficient outcomes. Key Applications (subject to a NDA; no further details available) • Conversational Agents for Swiss Insurance and Electricity Companies Developed custom AI chatbots, enhancing customer service and operational efficiency. • Decision Automation (Insurance Company) Created a solution automating up to 80% of high-level medical decisions and reducing costs. •Incident Analysis (Electricity Company) Implemented a system for analyzing unstructured data, improving incident management and predictability. Key Responsibilities • Comprehensive Project Management Oversee the entire lifecycle of AI projects, encompassing everything from initial design and planning to final implementation and maintenance including architecture, cost control, development stages, and performance evaluation. • AI Solutions Development and Integration Focus on creating scalable, industry-specific AI solutions that ensure high performance and reliability. Developing but also seamlessly integrating them into clients' existing systems (backend and frontend). • Innovation and Impactful Investment Lead the innovation and practical implementation, aligning with the latest industry trends while focusing on delivering projects that offer measurable success and optimized ROI. • Effective Communication and Team Collaboration Take charge of fostering a collaborative environment within the team including change management, alignment with evolving client needs, and data governance. Technical Skills • Az AI Foundry / Az ML • Az Webapps / Az Container Registry / Docker / FastAPI • Az Cloud Functions, Python for data preprocessing automation. • Az Doc Intelligence for OCR. • Az AI Search / Az Redis Cache for vector search. • LangChain for NLP, Langgraph / Pydantic AI / Autogen for agents • Copilot / Power Automate • Az Devops versioning / CI/CD pipelines.

Member of a team of Data engineers within the Statistics team dedicated to the maintenance and constant improvement of a data warehouse exploiting all the data generated by the daily activity of milions of users and having the purpose of providing studies to clients allowing them to manage their activity as well as extracting epidemiological reports. - End to end development of new features from raw data to iterative improvements via client feedback. - Solving business problems using several tools and data sources within an ETL process and impacting data analysis issues on the data collection process. - Development of interactive Power BI reports allowing clients to make in depth analysis. - Organizing and facilitating training sessions and support with the managers and analysts of client companies. - Management of sensitive medical data taking into account the associated anonymization issues.

• Skill improvement on SAP technologies, discovery of client business and the specifics of the existing technical solution. • Participation in the development of multiple functionalities in the customer solution: - Conducting of various tests in integration environments - Detection and correction of defects in collaboration with IT teams responsible for development. • Participation in the various stages of the deployment of a technical solution: - Collection of needs and identification of deviations via the holding of daily meetings with international leaders. - Gathering and interpretation of the user / system requirements in the design specifications with the deployment team. - Follow-up of the deployment schedule: change management, organization of informative meetings, preparation of stakeholders for the project. - Participation and supervision of the data migration and conversion process by the IT teams. • Creation, deployment and maintenance of a document management structure facilitating future deployments.

Within a team of researchers in artificial intelligence in charge of improving the industrial sorting of plastics for recycling, the objective was to develop a decision-making mechanism overcoming the imperfect nature of the data provided by current spectrophotometric methods. - Analysis of spectral data of unrecognized plastic waste during sorting and identification of the different types and sources of imperfections. - Research work and study of mathematical theories concerning fuzzy logic and other decision-making theories with imperfect data in order to determine the application potential. - Implementation of an algorithm based on fuzzy logic (RStudio) to improve the prediction and determination of the impact and limits of the method. - Documentation of the study and formatting of the results for publication.

Drawing up of the Quality Manual in order to obtain ISO 9001 certification. Setting up of an ERP (Oodoo).

Start up aiming to develop a virtual assistant improving the user experience and suggesting recipes on supermarket drive websites. - Market study and business model. - Drawing up of the functional specifications of the product. - Involvement in the design of the technical solution.