Maram Issaoui

Computer Engineering Program

Ibn Khaldoun University (UIK) · 2023 – Present

4th Year Engineering Student in AI and Data Science

AI/Data Science Engineering Intern — Gias Distribution

Jul – Aug 2026 · Tunisia · 2 months

Design of an intelligent decision-support solution to optimize sales representatives' visit routes, planning tours from commercial data according to revenue targets, business priorities and operational constraints.

• Design and development of a decision-support solution to optimize the visit circuits of sales representatives
• Exploitation of commercial data to plan tours according to revenue targets, business priorities and operational constraints
• Route optimization to maximize visit profitability, improve customer coverage and strengthen product promotion
• Field visits to customer sites to analyze business needs, validate operational constraints and collect the data required for the project
• Writing analysis and follow-up reports, and presenting results and recommendations to stakeholders

• Delivered an optimization engine aligning sales tours with revenue targets and operational constraints
• Presented analysis reports and recommendations to business stakeholders

Technologies: Python, Pandas, NumPy, LangChain, LangGraph, OSRM, Google OR-Tools, Streamlit, Plotly

Data Science Instructor — GOMYCODE

Mar 2026 – Present · Remote (North Africa) · Current

Teaches a 5-month, 200+ hour Data Science course covering machine learning and deep learning with Python (TensorFlow, PyTorch) to 10 adult students from different backgrounds and nationalities.

• Design and deliver practical training in data science, machine learning and deep learning with Python (TensorFlow, PyTorch)
• Accompany learners through end-to-end projects: data preparation, training, evaluation and interpretation of results
• Implement best practices for model validation, performance optimization and deployment principles
• Develop and supervise applied projects including preprocessing, experimentation, metrics tracking and simple inference APIs, with production-oriented code standards

• Helped learners build end-to-end AI and data science projects
• Applied production-oriented standards to training projects and inference APIs

Technologies: Python, TensorFlow, PyTorch, Machine Learning, Deep Learning, Data Analysis

AI Engineering Intern — Medin Fund Management

Oct – Dec 2025 · Remote · 3 months

AI-powered automation system for competitive analysis built with n8n.

• Built an AI-powered automation system for competitive analysis using n8n
• Implemented API integrations for data collection and processing
• Created automated workflows for generating insights and reports
• Enabled faster decision-making through real-time analytics dashboards

• Reduced analysis time by 60% through automation
• Streamlined decision-making process for stakeholders

Technologies: n8n, Python, APIs, Automation, Data Analysis

AI Engineering Intern — Sagemcom

Jul – Sep 2025 · Ezzahra, Tunisia · 3 months

Computer vision system for industrial quality control with real-time anomaly detection.

• Designed and developed a computer vision system for industrial quality control
• Implemented deep learning models for real-time anomaly detection
• Analyzed and optimized production images to improve precision, robustness and reliability
• Integrated and evaluated solutions in an industrial environment to reduce manufacturing errors and improve efficiency

• Achieved +92% precision in defect detection
• Reduced inference time by 18 ms
• Analyzed 2000+ production images

Technologies: Python, OpenCV, TensorFlow, Computer Vision, Deep Learning

AI Engineering Intern — Caredify

Jun – Aug 2025 · Remote · 3 months

Transformer architectures for automated anomaly detection in ECG data.

• Implemented transformer architectures for ECG anomaly detection
• Developed deep learning models for medical signal processing
• Created automated pipelines for real-time health monitoring
• Collaborated with healthcare professionals to validate results

• Achieved high accuracy in anomaly detection
• Contributed to healthcare innovation through AI

Technologies: Python, PyTorch, Transformers, ECG Processing, Medical AI

AI Engineering Intern — SESIT

Jul – Sep 2024 · Ariana, Tunisia · 3 months

GAN-based synthetic data generation and classification of local olive maturity stages.

• Developed an AI system based on GANs to generate synthetic data
• Trained deep learning models to classify olive maturity stages
• Preprocessed and optimized image data in a restricted-data context
• Improved performance through metrics analysis and architecture adjustments

• Achieved +98% classification accuracy
• Collected 1000+ local images

Technologies: Python, GANs, Computer Vision, Classification, Agricultural AI, Synthetic Data

Introductory Intern — BIAT

Jul – Aug 2023 · Tunis, Tunisia · 2 months

Observation and analysis of daily banking operations with customer relationship support.

• Observed and analyzed daily banking operations
• Participated in customer relationship tasks: reception, request processing and administrative support
• Documented workflows and procedures
• Supported analytics and reporting initiatives

• Gained valuable experience in the banking sector
• Contributed to process improvement documentation

Technologies: Excel, Data Analysis, Documentation, Customer Service

StrokeSight

StrokeSight is a comprehensive stroke diagnostic application that combines deep learning-based lesion detection with explainable AI techniques to provide interpretable results for medical professionals.

• Brain lesion detection using Transformer architectures
• Segmentation with UNet for precise localization
• Explainability through Grad-CAM visualization
• User-friendly interface for medical practitioners

• Achieved high accuracy in lesion detection
• Provided interpretable results for clinical decision-making

Technologies: Python, PyTorch, Transformers, UNet, Grad-CAM, Medical Imaging

Focus areas: Medical AI, Explainable AI, Computer Vision

XAI-CalloSeg

XAI-CalloSeg is an advanced medical imaging project that automates the segmentation of the corpus callosum from MRI scans using deep learning, with built-in explainability features.

• Automatic segmentation using UNet architecture
• High-precision localization of corpus callosum
• Explainability through Grad-CAM and Smooth Grad-CAM
• Vector Quantization for efficient representation

• Dice Score: 0.9981
• Accuracy: 0.9958
• Published at CoDIT 2025 conference

Technologies: Python, PyTorch, UNet, Grad-CAM, Smooth Grad-CAM, VQ

Focus areas: Medical AI, Explainable AI, Computer Vision

Published at CoDIT 2025

AI-MammoSaver

AI-MammoSaver is a breast cancer detection application that uses Convolutional Neural Networks with transfer learning to analyze mammography images and identify potential tumors, with automatic report generation via Gemini.

• CNN-based tumor detection from mammography images
• Transfer learning for improved performance
• Automatic report generation via Gemini
• User-friendly diagnostic interface

• High accuracy in tumor detection
• Reduced false positive rate through optimization

Technologies: Python, TensorFlow, CNN, Transfer Learning, Gemini, Medical Imaging

Focus areas: Medical AI, Computer Vision, Generative AI

Published at ESANN 2026

CFTRaX

CFTRaX is a rare genetic variant prioritization system that combines deep learning, uncertainty modeling and multi-model consensus with an explainability layer and GenAI under European ethical constraints.

• Deep learning-based variant scoring and ranking
• Uncertainty modeling and multi-model consensus
• Explainable AI layer for clinical interpretability
• GenAI integration constrained by European ethical guidelines

• Built a complete prioritization pipeline for rare genetic variants
• Integrated explainability and GenAI under ethical constraints

Technologies: Python, RankNet, ListNet, Llama 3, CFTR, Deep Learning

Focus areas: Bioinformatics, Explainable AI, Data Science, Generative AI

ProductPilot AI

ProductPilot AI is a Product Management platform powered by multi-agent systems and LLMs to automate market analysis, feature prioritization and product decision-making.

• Multi-agent architecture for market analysis automation
• LLM-driven feature prioritization and decision support
• Workflow orchestration for product management tasks
• Generative AI pipeline for product insights

• Automated market analysis and feature prioritization workflows
• Demonstrated agentic AI for product decision support

Technologies: Python, LangGraph, LLM, Agentic AI, Multi-Agent Systems, NLP, Workflow Orchestration

Focus areas: Product Management, NLP, Agentic AI

Booklytics

Booklytics is a hotel analytics dashboard built with Dash and Plotly to track prices, ratings, seasonal trends and detect overvalued hotels from web-scraped data.

• Interactive dashboards for hotel price and rating tracking
• Seasonal trend analysis and market insights
• Overvalued-hotel detection from scraped web data
• Business intelligence visualizations

• Built an end-to-end BI dashboard for hospitality data
• Enabled data-driven pricing and market analysis

Technologies: Python, Dash, Plotly, Selenium, Web Scraping, Business Intelligence

Focus areas: Business Intelligence, Web Scraping, Data Science

XAI-Enabled Custom CNN for Cross-Modal Generalization in Breast Cancer Detection

ESANN 2026 — European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

Bruges, Belgium (and online) · 22–24 April 2026

Authors: Maram Issaoui, Amal Jlassi, Abir Baâzaoui, Walid Barhoumi

Accepted paper on explainable AI for breast cancer detection across multiple imaging modalities.

• A single fine-tuned CNN generalizes across both mammography and histopathology under one cross-modal protocol.
• Stable, high accuracy on both imaging types — robust to domain shifts and heterogeneous clinical conditions.
• Combines model-agnostic (LIME, SHAP) and model-specific (Grad-CAM) explainability for a performance–interpretability balance.
• Delivers clinically meaningful visual and feature-level insights for transparent diagnostic decisions.

Technologies: Custom CNN, Transfer Learning, Grad-CAM, LIME, SHAP

Keywords: Explainable AI, Breast Cancer Detection, Cross-Modal Generalization, Mammography, Histopathology

Explainable AI-Driven Prediction of Chemoresistance in Cellular Models

ISSATK 2026 — IEEE International Symposium of Systems, Advanced Technologies and Knowledge

Hammamet, Tunisia · 25–27 April 2026

Authors: Maram Issaoui, Amal Jlassi, Boudour Ammar, Ichraf Jbir

Accepted paper on explainable AI for predicting chemoresistance in cellular models.

• Integrated pipeline: XGBoost + single-cell inference + explainable AI + automated reporting.
• Captures subtle drug-response differences across cancer subtypes from z-scored data.
• Single-cell RNA-seq inference reveals resistant subpopulations and cell-specific drug effects.
• SHAP explains feature contributions; an LLM turns complex outputs into clinician-friendly reports.

Technologies: XGBoost, SHAP, scRNA-seq Inference, Large Language Models

Keywords: Precision Oncology, Chemoresistance, Explainable AI, Single-Cell Analysis, Generative AI

Vision Transformer-Assisted Defect Classification and YOLOv8-Based Localization in Industrial Quality Control

IEEE IMC-SSGP 2025 — 2nd International Multi-Conference on Smart Systems & Green Process

Hammamet, Tunisia · 30 October – 2 November 2025

Authors: Arwa Belhedi, Maram Issaoui, Ahmed Khalil Haous, Raef Cherif

Research publication at the IEEE International Multi-Conference on Smart Systems & Green Process.

• A Doosan collaborative robot with a 2.5D camera works alongside AI models for real-time inspection.
• Vision Transformer reached 95.1% accuracy for defect classification.
• YOLOv8 reached 95.8% accuracy for defect localization, outperforming lightweight CNNs.
• Trained on 1,050 annotated images; enables automated sorting on production lines for Industry 5.0.

Technologies: Vision Transformer (ViT), YOLOv8, MobileNetV2, EfficientNet, Doosan Cobot, 2.5D Camera

Keywords: Industry 5.0, Quality Control, Defect Detection, Collaborative Robotics, Computer Vision

A Vector Quantization-Based U-Net for Robust Segmentation of Corpus Callosum

CoDIT 2025 — 11th International Conference on Control, Decision and Information Technologies

Split, Croatia · 15–18 July 2025

Authors: Amal Jlassi, Maram Issaoui, Sami Hafsi, Ezequiel de la Rosa, Ahmed Harbaoui

Published article on explainable AI for corpus callosum segmentation.

• A novel VQ-UNet adds a Vector Quantization memory module inside the U-Net bottleneck.
• Richer feature representation cuts the model's dependency on large annotated datasets.
• Saliency maps add interpretability for trustworthy volumetric quantification of the corpus callosum.
• Outperforms state-of-the-art with up to 2% higher Dice scores than U-Net variants.

Technologies: U-Net, Vector Quantization, Saliency Maps, Brain MRI

Keywords: Medical Image Segmentation, Corpus Callosum, Explainable AI, Brain MRI

Build with AI: Masr Edition 2026 — Completed

2026

Successfully completed the Build with AI: Masr Edition 2026 program by Google Developer Groups (GDG), supported by the Information Technology Institute (ITI). Selected among 5,000 developers, the program covered Cloud Computing, Generative AI, multi-agent systems with LangGraph, and real-world AI product deployment.

2025 SPARK Academy — Completed

February – October 2025

Successfully completed the 2025 SPARK Academy (Sprint AI Training for African Medical Imaging Knowledge Translation), a hybrid program from February to October 2025 organized by CAMERA. The program covered AI, medical imaging, and deep learning fundamentals to drive healthcare innovation and accessibility across Africa.

MICCAI Summer School 2025 — Attended

14–18 July 2025

Attended the RISE-MICCAI Summer School 2025 (14–18 July), a week of cutting-edge lectures and hands-on workshops on AI for medical imaging. Highlights included Diffusion Models, Graph Learning, Agentic AI for Medicine, Scientific Writing, Uncertainty Quantification, and an inspirational talk by Udunna Anazodo.

BIGTECH AFRICA Poster Session — Exhibitor

2025

Presented my work at the BIGTECH AFRICA poster session, organized by the Tunisian AI Society.

AI4Blue Horizons Hackathon 2025 — 1st Place

6 February 2025

Won 1st place at AI4Blue Horizons with AgriFlow, a project leveraging AI to optimize hydroponic systems and reduce water consumption by 90%. The competition was organized by CJD Tunis Horizon and UIK - Université Ibn Khaldoun. I also discussed AgriFlow in an Express FM interview about entrepreneurship and sustainable water use.

DEEPTECH WINTER ACADEMY — Organizer

December 2024

Organized the DEEPTECH WINTER ACADEMY as President of ATIA Club UIK, featuring insightful conferences with guest speakers on technology and innovation. Also earned the NVIDIA Deep Learning Fundamentals Certificate, reflecting continued dedication to AI and deep learning.

BrainX Program Selection — Selected

2024

Best Multidisciplinary Integration Project 2024 — 1st Prize

22 May 2024

My Multidisciplinary Integration Project 'XAI-CalloSeg' was selected for the prestigious 'Boom Of M.I.P' at UIK - Université Ibn Khaldoun and recognized as Best Project.

17th Carthage Insurance Hackathon 2024 — 1st Place

2024

Our team Insure Genius won first place in the fraud detection challenge at the 17th Carthage Insurance RDV, organized by FTUSA (Fédération Tunisienne des Sociétés d'Assurances) and ATIA TUNISIE. We were awarded a 5,000 TND prize.

IEEE AMCAI Conference — Participant

13–15 December 2023

Participated in the inaugural IEEE Afro-Mediterranean Conference on Artificial Intelligence (AMCAI), organized by ATIA TUNISIE and sponsored by the IEEE Africa Council. The conference showcased peer-reviewed research in applied AI through industrial panels on Industry 4.0 and cybersecurity, and keynotes by global AI authorities including Professor Timothy Jung, who shared insights on integrating AI into XR.

Manouba Networking Day 2023 — Exhibitor

2023

Exhibited at the 5th edition of Manouba Networking Day, representing ATIA Club UIK under the theme "Artificial Intelligence & Socio-Economic Prosperity: Alliance for a Sustainable Future." The event, held in partnership with ATIA TUNISIE and Association Alumni ISCAE, brought together students, professionals, and organizations to explore how AI can drive sustainable socio-economic development.

Vice Chair — IEEE UIK SB

2025 – 2026

Helped lead the IEEE Student Branch at UIK, organizing technical events and building partnerships with academic and industry collaborators.

• Co-organized technical events and student activities.
• Collaborated with external partners and sponsors.
• Supported the branch's growth and member engagement.

President — ATIA UIK Club

2023 – 2024

Led the university's AI club, coordinating activities and driving flagship events that promoted AI literacy and peer learning on campus.

• Organized and ran the DeepTech Winter Academy, featuring guest speakers on technology and innovation.
• Coordinated the club's activities and events throughout the year.
• Promoted AI literacy through peer-learning initiatives.

IT Member → General Secretary — JCI Bardo

2023 – 2025

Progressed through leadership roles at Junior Chamber International Bardo, from IT and project work to running the chapter's administration and coordination.

• IT team member (2023) — helped develop the JCI Bardo website.
• Co-Director of the Xsebhom project (2024) — led project development and team coordination.
• General Secretary (2025) — managed administrative tasks and managerial coordination.