[nom]
AI Engineer & Data Scientist
Tel : [téléphone] · || · Localisation :Casablanca, MOROCCO
Email : [e-mail] Linkedin : [lien] | GitHub : [lien]
EXPERIENCE
AI Engineer Intern - Agent Workflow Automation (n8n) · August 2025– September 2025
Elavi - [lien] · · · · · · · · · · · · REMOTE , Morocco
● · Designed an intelligent interface based on an LLM agent, capable of automating customer communication and providing fast, relevant responses.
● · Deployed a messaging workflow for sending messages via email and LinkedIn. This workflow collects client information, including
previous clients from LinkedIn, and handles the detection and correction of invalid emails using the client’s first name, last name, and domain, leveraging Apollo (people search, company enrichment…), Apify (Google scraper, LinkedIn profile scraper…), Phantombuster, Findymail, and LLM Agents. and Implemented a centralized database to store conversation history, enabling the agent to generate personalized responses based on previous interactions.
● · Developed and integrated Python open-source APIs for email verification (valid or invalid) and professional address lookup based
on first name, last name, and domain, deployed via Docker and Docker Hub to ensure portability and scalability.
AI Engineer Intern - Handwriting Recognition with Neural Networks · Feb 2025– Jul 2025
DXC Technology · · · · · · · · · Technopolis ,Sala Al Jadida, Morocco
● · Developed a handwriting recognition system capable of converting handwritten text into digital format using neural networks.Designed and
implemented two model architectures.A hybrid CNN + BiLSTM + CTC(Connectionist Temporal Classification) model, using type ResNet v1/v2 as backbone for feature extraction, enhanced with Self-Attention and Multi-Head Attention modules.A Transformer-based encoder-decoder architecture integrated with ResNet-50/101 backbones, enabling comparative analysis on robustness and accuracy for complex paragraph-level sequence modeling.
● · Evaluated models with CER (Character Error Rate), WER (Word Error Rate), SER (Sentence Error Rate), CTC Loss, and Accuracy for
thorough performance assessment.and Employed IAM and RIMES datasets for diverse handwriting training samples.
● · Integrated local LLMs with external APIs using LangChain and FastAPI (including Groq) to improve syntactic and semantic quality of transcribed
text; securely managed API keys with dotenv and utilized advanced training callbacks (EarlyStopping, ModelCheckpoint, TensorBoard, ReduceLROnPlateau, custom TrainLogger) to monitor metrics, visualize progress, and optimize learning rate scheduling.
● · Applied advanced segmentation techniques for paragraph and line detection using a heatmap-based approach to accurately identify text areas and
segment lines. This robust method also included automatic preprocessing steps such as correcting image and line slant, denoising, and contrast enhancement, enabling the system to reliably recognize handwritten text even when images are distorted or captured from challenging angles or varying distances.
● · Developed a user-friendly web interface allowing users to easily upload or drag-and-drop images for real-time prediction. The entire application was
fully containerized and deployed using Docker for efficient and portable deployment.The application takes between 7 and 22 seconds .
AI Engineer Intern · Ap 2024 – Jun 2024
IBN TOFAIL UNIVERSITÉ - SETIME LAB · · · · · · · · · Kenitra, Morocco
● · Developed a real-time obstacle detection system with 91.9% accuracy using fine-tuned MobileNetV2, 81% with a custom CNN, and integrated an
auditory alert system using Node-RED for navigation assistance and built an API with Flask to process image and video inputs, enabling real-time predictions and audio alerts.implemented video frame processing with FFmpeg for real-time obstacle detection on both images and videos.
● · Enhanced overall system efficiency and responsiveness significantly to ensure low-latency obstacle detection and immediate audio feedback for users.
PROJECT
Custom-OCR-YOLO || Python, OpenCV, YOLOv3, Tesseract OCR, Pandas, Numpy, CSV
● · Developed a custom OCR solution by integrating YOLOv3 for object detection and Tesseract for text extraction, enabling automated
reading of lab report images. Implemented robust image preprocessing (cropping, resizing, thresholding) using OpenCV and Numpy to optimize recognition accuracy..
● · Automated data extraction and conversion to editable CSV files with Pandas, streamlining information management for laboratory results.
Bank IAV API || Flask, JSON Web Token (JWT), PostgreSQL,Postman API, jira ,Scrum ,Agile,SQLAlchemy , gRPC
● · Developed a robust and scalable backend API for banking applications using Flask and gRPC, providing essential functionalities such as
user registration, authentication, and transaction management.Integrated SQLAlchemy as the ORM(Object-Relational Mapping) to manage complex database operations efficiently with PostgreSQL, ensuring data integrity and scalability in financial transactions.
● · Implemented secure JWT-based authentication using Flask-JWT-Extended and integrated the system with Jira to prioritize development tasks.
EDUCATION
M.S in Artificial Intelligence and Virtual Reality · 2023 - 2025
IBN TOFAIL UNIVERSITY · Kenitra, MOROCCO
● · Machine Learning, Deep Learning, Reinforcement Learning, Natural Language Processing, Computer vision, Agent Workflow n8n
B.S. in Mathematics and Computer Science, Major in Computer Science · 2020 - 2023
Ibn Zohr University · Ouarzazate, MOROCCO
● · Data Structures , Algorithms , Operating Systems , Computer Architecture , Databases, Linear Algebra & Statistics , Linux User
Baccalaureate in Physical Sciences · 2019- 2020
Sidi Amrou High School · Tazarine , Zagora, MOROCCO
SKILLS
Technical Skills : Hugging Face, RAG, GitHub, Git, Linux, Jira,Postman ,Swagger,,Bash,SSH, Docker, Kubernetes. LangChain,LlamaIndex,Flask,,FastAPI , gRP,Streamlit,ML0ps,LangGraph,LangSmith,Workflow n8n ,Ultralytics ,PaddleOCR,Trandformers,(ViT) Vision Transformer, RT-DETR , LightGlue , Qwen-2VL , YOLO
, PaddleLayout, D-FINE ,LayoutLMv3 , Donut (Document Understanding Transformer) , RT-DETR(Real-Time DEtection TRansformer) ,Tesseract OCR ,Vertical Attention
Databases: MySQL, SQL Server, SQLAlchemy, PostgreSQL, MongoDB,Cassandra,, Faiss, ChromaDB,DuckDB,Redis
Area of Expertise: Software & Data Engineering, AI Engineering, Data Science and Statistics, Computer Vision , Organization, Linux User & DevOps
Certifications: Intro to LLMs, Machine Learning in Python, Intro to NLP for AI, Advanced SQL
Languages Spoken: Arabic (Native), English (Intermediate), French (professionnelle)
