ABDELMAJID ZADDI

I work on end-to-end solution delivery — from system design and architecture to implementation, testing, and deployment. My work involves building scalable backend services using Python (FastAPI), Node.js, Java (Spring Boot), and Go, and deploying them with Docker, Kubernetes, GitHub Actions, and Jenkins. I also develop AI-driven solutions including RAG pipelines, agentic workflows, graph-based RAG, deep learning models, and OCR systems.

I designed and developed a fully functional AI-powered trading bot leveraging LSTM (Long Short-Term Memory) neural networks to analyze financial time series data and generate real-time trading signals with high accuracy. The solution covers the entire pipeline — from data collection and preprocessing to model training, prediction, and automated execution. The project included building a scalable backend using Python (FastAPI/Django), integrating machine learning and deep learning models for predictive analytics, and creating a user-friendly frontend dashboard with JavaScript/React for live monitoring and performance visualization. I also implemented DevOps workflows and cloud deployment to ensure continuous delivery and scalability

As an AI Developer and Software Engineer at Cash Plus, I designed and implemented an advanced multilingual chatbot powered by RAG (Retrieval-Augmented Generation) and agentic RAG. The chatbot is capable of answering customer inquiries across multiple domains and performing actions such as creating claims, handling reclamations, and automating support tasks. I was responsible for the end-to-end development of the solution — from system design and backend implementation to integration and deployment. The project was built using Python (Flask, Django) for the backend, JavaScript for the frontend, and integrated with LLMs and vector databases to ensure accurate, context-aware responses.

Text summarization is one of the most interesting problems in NLP. It’s hard for us, as humans, to manually extract the summary of a large document of text. To solve this problem, we use automatic text summarization. It’s a way of identifying meaningful information in a document and summarizing it while conserving the overall meaning. The purpose is to present a shorter version of the original text while preserving the semantics.

I designed and developed a complete Neural Machine Translation (NMT) system from scratch to translate between Darija (Moroccan Arabic) and English, a low-resource language pair with limited publicly available data. The project involved the entire pipeline, including data collection, text preprocessing, tokenization, custom dataset creation, model architecture design, training, evaluation, and deployment. I implemented deep learning-based sequence-to-sequence models with attention and transformer architectures to achieve high translation accuracy and natural language fluency. The backend was built with Python (PyTorch/TensorFlow), and I integrated the model into a simple web application using FastAPI/Flask for serving translations in real time. This project demonstrates my ability to build end-to-end NLP systems tailored to low-resource languages and apply cutting-edge language model techniques to solve real-world communication challenges