Ben Jaballah Mohamed

Development of an Intelligent Assistant Chatbot (RAG + AI Agents) for Saas • Designed and implemented a Retrieval-Augmented Generation (RAG) architecture combining contextual document search with text generation, integrated with internal business APIs to deliver accurate, real-time, and context-aware answers. • Built and orchestrated AI agents (LangChain + LangGraph) capable of combining document retrieval, API calls, analytics, and automated actions. Automatically generated tools from OpenAPI documentation to enable seamless system interactions. • Applied advanced Prompt Engineering for language detection, tone adjustment, source prioritization, and fine-grained access control. • Deployed the solution on Azure AI Foundry and the tools on MCP servers, ensuring scalability, security, high availability, and seamless SaaS integration. Tech Stack • Vector Databases: ChromaDB, Pinecone, Weaviate • LLMs: OpenAI (GPT-4, GPT-4o), Mistral (Small, Large), Claude, Hugging Face models • Frameworks & Orchestration: LangChain, LangGraph • Embeddings: OpenAI (text-embedding-ada-002, text-embedding-3-large), Mistral Embed, Hugging Face • Cloud & Infrastructure: Azure AI Foundry, MCP servers, Docker

Development of an AI engine integrated into the KPAX SaaS platform to anticipate toner replacement • Designed and implemented predictive models to forecast toner usage and replacement needs, enabling proactive maintenance and reducing downtime. • Collected, cleaned, and prepared multi-source data (device logs, IoT signals, databases) to build robust training datasets. • Trained and validated time-series models (Random Forest and deep learning approaches with TensorFlow) to optimize prediction accuracy. • Deployed the AI engine via secure FastAPI services, ensuring real-time access to predictions and seamless integration with the SaaS platform. • Implemented MLOps best practices (automation, monitoring, governance) using Dataiku pipelines and Azure AI Foundry for scalability and reliability. • Containerized the solution with Docker for portability and multi-environment deployment. Tech Stack • Programming: Python • APIs: FastAPI (secure real-time services) • Machine Learning: TensorFlow, scikit-learn, Random Forest , Arima , SArima • MLOps & Data: Dataiku, Azure AI Foundry • Infrastructure: Docker

Presentations and training for print-market professionals and IT service companies across French-speaking Africa • Delivered multiple presentations and training sessions enabling professionals to unlock new revenue streams through advanced print technologies. • Covered topics including centralized fleet management, print-workflow optimization, document capture/distribution, and digital document creation. • Equipped Naxxum Solutions’ partners and clients with the expertise to streamline workflows, reduce costs, and enhance document management. • Contributed directly to customer value creation and supported the company’s overall business growth

Led a 62-member student team at Enactus ISI, driving social impact projects through entrepreneurial action. Defined vision and annual plan, fostered innovation, and ensured smooth operations through effective delegation and mentorship. Oversaw project execution from planning to delivery, proactively managing risks and resolving conflicts. Strengthened leadership, communication, and project-management skills while inspiring young talent to create positive change.