Achraf Laajimi

Conception architecturale d’un assistant IA d’entreprise évoluant d’un RAG classique vers une recherche hybride combinant la vectorisation sous Qdrant et des requêtes SQL PostgreSQL en temps réel. Déploiement d’un tableau de bord d’administration pour le monitoring des performances, la gestion dynamique de la configuration et le suivi budgétaire strict des consommations LLM. Sécurisation zéro-confiance de l’application intégrant une authentification JWT, la neutralisation du prompt injection et la détection des fuites d’instructions par mécanisme canari.

Design and development of an intelligent system based on an advanced multi-agent architecture. A first module, the exam agent, automates the analysis and grading of various types of exams through state orchestration built with LangGraph. This module integrates computer vision techniques based on OpenCV, as well as a hybrid text extraction mechanism combining OCR and VLM models. Concurrently, a conversational agent assists parents in the personalized academic monitoring of their children, relying on an asynchronous task planner based on a DAG. It is capable of generating adaptive responses in different formats, including tables and visualizations via QuickChart. Interoperability between these AI agents, institutional databases, and external APIs is standardized via an MCP gateway. Overall security and control of the context of exchanges are ensured by Casbin, combining RBAC and ABAC models. System observability and performance monitoring are ensured by Langfuse.

Developed a context-aware MCP server for integrating AI agents with enterprise backends. Implemented goal-based permission filtering backed by MongoDB, an in-memory LRU conversation cache for session continuity, and RAG integration with an external vector database and LLM for context retrieval and response generation. Supports SSE streaming and REST endpoints, async processing, TTL caching and failover behavior.

I've developed a comprehensive AI-powered vocal agent that combines cutting-edge technologies to deliver a seamless user experience. GRACE features natural language processing (NLP) capabilities for intelligent voice interactions, web scraping functionality for real-time data acquisition, and integrates large language models (LLMs) for contextual conversations. The platform includes a sophisticated resume analyzer that provides ATS scoring and detailed feedback to help job seekers optimize their applications. Additionally, I've implemented advanced image generation and editing capabilities, including AI-powered packshot creation, shadow effects, lifestyle shot generation, generative fill, and intelligent foreground removal. Built with modern web technologies (JavaScript, Python, Bootstrap) and featuring a sleek neon-themed UI with animated loading states, this project demonstrates the power of combining multiple AI technologies into a unified, user-friendly platform. The system showcases real-world applications of AI in both professional development and creative workflows.

I’ve been developing the Project Management and Time Attendance system, a robust solution to streamline project oversight and track attendance effectively. Powered by a Node.js/Express backend with JWT authentication, Socket.IO for real-time notifications, and Swagger documentation, alongside a Flutter mobile app, this project offers seamless task management, product inventory tracking, and precise time attendance features. This showcases my skills in full-stack development and mobile app creation!

As part of my internship, I participated in the development of a Customer Relationship Management (CRM) web application using the MERN stack (MongoDB, Express.js, React, Node.js). I designed and implemented key features such as user management, order tracking, sales analysis, real-time statistics, and interactive dashboards. The application provides an intuitive and responsive interface, enabling administrators to efficiently monitor business activities and improve customer relationship management. This project allowed me to strengthen my full-stack development skills by working on both the front-end and the back-end.