Ahmed Channoufi — Software Engineer

Who I am

About Me

I am a graduate in Software Engineering and Information Systems from ISLAIB (Institut Supérieur des Langues Appliquées et de l'Informatique de Béja). I have a versatile profile: I cover full-stack development, Machine Learning, databases, as well as a good understanding of Linux administration and Git.

I thrive at the intersection of software engineering and data science, turning complex problems into elegant, data-driven solutions. Trilingual in Arabic, French, and English, I bring both technical depth and cross-cultural communication to every project.

3+ Projects Shipped

3 Languages Spoken

Certifications

🐍

Data Science Essentials with Python Cisco Networking Academy · Sep 2026

View Certificate ↗

🖥️

Presenting Data HP LIFE · April 2026

View Certificate ↗

🗄️

SQL 365 Data Science · Nov 2024

View Certificate ↗

📐

Mathematics 365 Data Science · Dec 2024

View Certificate ↗

What I know

Technical Skills

💻 Languages

Python Java C# JavaScript HTML CSS PHP

⚙️ Frameworks & APIs

FastAPI Flask ASP.NET MVC REST APIs Chart.js

🤖 ML & Data Science

Scikit-learn XGBoost Pandas NumPy

🗄️ Databases

MySQL PostgreSQL Oracle SQL Server

🛠️ Tools & Systems

Linux Git GitHub

🌍 Languages Spoken

ArabicNative

FrenchIntermediate

EnglishGood Level

My journey

Experience & Education

Internship Feb 2026 – Jun 2026

Final-Year Project Internship (PFE)

Tunisie Telecom · Tunis, Tunisia

Designed a complete decision-support system following the CRISP-DM methodology to identify customers at risk of churn.
Handled data quality issues (missing values, duplicates, class imbalance); performed feature engineering and data visualization.
Trained and evaluated multiple machine learning models; selected XGBoost as the final model based on its performance, achieving 96% accuracy, 83% precision, 93% recall, and 88% F1-score on the churn class.
Validated the final model on real customer data, confirming strong and stable performance.
Built a Flask web application with an interactive dashboard and dynamic KPIs for prediction visualization.

PythonXGBoostSMOTEFlaskCRISP-DM

Internship June 2025

.NET Developer (PFA)

Asteelflash Tunisia · Soukra, Ariana, Tunisia

Developed a role-based (Administrator, User, Operator) ASP.NET MVC web application for industrial quality test management, implementing a validation workflow (Initial Test → Quality Control → Client Test) with strict business rules.
Designed the system architecture with Entity Framework and SQL Server, secure BCrypt authentication, and automatic test-type suggestions based on product history.
Built advanced search functionality, Chart.js analytics, and Excel export (EPPlus) for quality reporting.

C#ASP.NET MVCSQL ServerChart.jsEPPlus

Education 2023 – 2026

Bachelor's in Computer Science

ISLAIB · Software Engineering & Information Systems

Graduated with Honors 🎓

Education 2022 – 2023

Baccalaureate in Computer Science

Lycée Omrane Supérieur · Tunis

What I've built

Featured Projects

🤖

PFE · Internship @ Tunisie Telecom

Système Décisionnel de Prédiction du Churn Client (ML)

End-to-end Machine Learning decision-support platform built for Tunisie Telecom following the CRISP-DM methodology. The target file combined basic customer profile information along with the target variable (customer status for the following month), supplemented by 5 behavioral data files covering 12 months (outgoing/incoming calls, internet consumption, recharges, USSD). Cleaned and consolidated these files around the customer ID to construct 46 behavioral features (trend deltas, active months, recency). Balanced dataset with SMOTE and deployed in an interactive Flask Web Dashboard with automated multi-file ingestion, dynamic KPI analytics, and CSV reporting.

📊 Dataset Class Distribution & Preprocessing

🟢 Non-Churn (Class 0)

84.6% (23,009)

🔴 Churn (Class 1)

15.4% (4,190)

🧪 Model Comparison & Overfitting Analysis (8 Algorithms Evaluated)

Model	Accuracy	Precision	Recall	F1-Score	Overfitting Risk (Train/Test Gap)
Régression Logistique	91.0%	63.0%	95.0%	76.0%	Low (0.8%)
Naive Bayes	81.0%	45.0%	96.0%	61.0%	Low (1.1%)
Arbre de Décision	91.0%	64.0%	91.0%	75.0%	Moderate (3.2%)
K-NN (K-Nearest Neighbors)	92.0%	66.0%	95.0%	77.0%	Low (1.2%)
SVM Linéaire	91.0%	63.0%	95.0%	76.0%	Low (0.9%)
Random Forest	95.0%	81.0%	93.0%	87.0%	⚠️ High (5.8% Gap — Overfitting Risk)
Gradient Boosting	96.0%	82.0%	93.0%	87.0%	Low (1.5% Gap — Slow 12s Train Time)
⭐ XGBoost (Selected Model)	96.0%	83%	93%	88%	✅ Optimal (1.46% Gap — No Overfitting, 0.93s)

✅ Production Validation on Unseen Real Data (34,000 New Customers)

96%Accuracy

82%Precision

98%Recall (4,906/5,021 Caught)

89%F1-Score

🔒 Model trained in 0.93s with 1.46% generalization gap, catching 4,906 true churners out of 5,021 in new validation dataset.

🌐 Flask Decision-Support Web Application Features

1

Automated Ingestion 6-File monthly upload (Calls, Data, Recharges, USSD)

➔

2

Real-Time Pipeline Auto-cleaning, 46-feature extraction & XGBoost inference

➔

3

KPI & Export Dashboard Risk scorecards, distribution charts & 1-click CSV export

📊 Features interactive KPI scorecards, a risk-level pie chart based on a 3-tier probability threshold (Low: 0–30%, Medium: 30–70%, High: 70–100%), a probability distribution histogram, and a sorted contract risk list.

PythonXGBoostScikit-learnSMOTEFlaskPandasNumPyMin-Max ScalingCRISP-DM

🔬

PFA · Internship @ Asteelflash

Application Web de Gestion des Tests Industriels (EMS)

Industrial Web Platform developed for Asteelflash (Electronic Manufacturing Services - EMS). Implements an end-to-end 3-level quality control workflow for electronic boards, complete with strict business rejection rules, real-time hardware assignment, multi-role security (RBAC), live Chart.js analytics, and automated EPPlus Excel report generation.

🔍 Understanding the 3 Test Levels

1

Initial Test Initial Test — Checks that the electronic board's identifier (serial number) is correctly structured and valid upon registration in the system.

➔

2

Quality Control Quality Control — Physical inspection of the board to confirm it's in good condition (no visible or hardware defects) before moving forward.

➔

3

Client Test (Level 3) Client Test — Verifies that the board is correctly programmed according to the specific requirements and configuration requested by the end client.

⛔ Rejection Rule: Automatic product rejection upon 2 consecutive failures at any level. Full audit trail recorded.

🛡️ Smart System Features

Automatic next-test suggestion — After each successful test, the system automatically suggests the next test to perform based on the board's history.
Error prevention — If the user attempts an incorrect action (wrong order, invalid step), the system blocks it and stops the process immediately.
Smart filtering — When running a test, the operator can filter by machine and operator to quickly find the correct board/line to test.

👥 Multi-Role Authorization & Key Capabilities

Role	Key Responsibilities	Special Features
Administrator	User management, global test history, system-wide KPIs	Chart.js category analytics & Excel export
User	Test registration, product history inspection, advanced search	Multi-criteria filtering (Dates, Boards, Machines)
Technician	Executes Initial Test and Quality Control	60-second countdown double confirmation
Engineer	Executes Client Test	60-second countdown double confirmation
Supervisor	Executes all test types	60-second countdown double confirmation

🏭 Operational Performance & Output Metrics

3Validation Stages

5RBAC Roles

3 minAvg Test Cycle

100%Traceability

🔒 Integrated with BCrypt security, EF ORM, SQL Server, and EPPlus for instant production Excel exports.

C#ASP.NET MVCEntity FrameworkSQL ServerBCryptChart.jsEPPlusStarUMLCascade Model

🏪

Full-Stack · Personal Project

Magasin+ — Store Management System

Full-featured point-of-sale web application with fast checkout, smart stock management with low-stock alerts, product categories, cashier roles (cashier/supervisor), sales reports with CSV export, printable receipts, and JWT-secured authentication.

FastAPISQLModelPostgreSQLJWTPython

📚

Backend · Academic Project

Biblivre REST API

A complete library management REST API built in Java. Handles books, members, loans, and returns with full CRUD operations and clean RESTful architecture.

JavaREST API

View All on GitHub

Let's connect

Get In Touch

Have a project in mind or want to hire me? I'd love to hear from you!

Email channoufiahmed257@gmail.com

Phone +216 58 997 620

LinkedIn ahmed-channoufi

GitHub AhmedChannoufii