Yosr Fourati

Software Engineer

Software Engineer MS Graduate and Researcher building intelligent systems at the intersection of software engineering and large language models. Published IEEE author. Full-stack engineer with production experience.

⬡ GitHub in LinkedIn 🎓 Google Scholar
1 IEEE Publication
3.96 Graduate GPA · Oakland University
2× Best Student Project
5+ Production Projects

// who I am

About Me

I'm a software engineer and AI/ML researcher. My work sits at the intersection of backend systems, full-stack development, and applied machine learning.

During my time as a Research Assistant at Oakland University, I co-authored CARGO, a lightweight LLM routing framework published at IEEE CASCON 2025. The framework dynamically selects the most suitable language model per prompt, achieving 76.4% top-1 accuracy across GPT-4o, Claude 3.5 Sonnet, DeepSeek V3, and Perplexity Sonar.

On the engineering side, I've shipped production features across e-commerce platforms, real estate apps, and banking systems using Java, Spring Boot, Angular, Next.js, and Python. I care about clean architecture, maintainable code, and systems that actually work.

🔬

AI/ML Research

LLM routing, ensemble model selection, multi-agent pipelines. Published at IEEE CASCON 2025.

⚙️

Full-Stack Engineering

Spring Boot, Angular, Next.js, Node.js : from REST APIs to responsive frontends.

☁️

Cloud & DevOps

AWS (EC2, S3, RDS, IAM), Docker, CI/CD pipelines, and Linux environments.

🏆

Academic Excellence

Two-time Best Student Project Award. President's List at Oakland University (4.0 GPA).

// where I studied

Education

MS in Software Engineering & Information Technology
Oakland University · Rochester, MI, USA
GPA: 3.96  |  President's List (4.0 in three semesters)
Machine Learning Natural Language Processing Cloud Computing Software Security Algorithms Info Retrieval & Knowledge Discovery Software Verification & Testing IT Project Management Independent Study (AI) Advanced Web Design Application
Sep 2024 – Apr 2026
🏅 President's List
BS in Software Engineering
Mediterranean Institute of Technology (MedTech) · Tunis, Tunisia
Capstone thesis: "Ensemble Model Selection for LLMs" supervised by Dr. Dorra Louati & Dr. Amine Barrak
Best Senior Student Project 🏆 Best Junior Student Project 🏆
Sep 2021 – May 2024
2× Award Winner

// published work

Research & Publications

My research focuses on making large language model deployments smarter and more efficient through dynamic routing, ensemble selection, and multi-agent coordination. Below are my published papers, thesis, and contributed research projects.

Research Assistant
Oakland University · Department of Computer Science & Engineering
September 2024 – August 2025
📄 IEEE CASCON 2025  ·  arXiv:2509.14899  ·  Peer-Reviewed Conference Paper
CARGO: A Framework for Confidence-Aware Routing of Large Language Models
A. Barrak, Y. Fourati, M. Olchawa, E. Ksontini, K. Zoghlami  ·  Oakland University & Mediterranean Institute of Technology
CARGO is a lightweight, confidence-aware framework for dynamic LLM selection. It employs a single embedding-based regressor trained on LLM-judged pairwise comparisons to predict model performance, with an optional binary classifier invoked when predictions are uncertain. Evaluated on GPT-4o, Claude 3.5 Sonnet, DeepSeek V3, and Perplexity Sonar achieving 76.4% top-1 accuracy and 72–89% win rates against individual models.
🔗 arXiv Published
📄 arXiv:2510.07614  ·  Contributed Research
Traceability and Accountability in Role-Specialized Multi-Agent LLM Pipelines
Amine Barrak et al.  ·  Contributor: Yosr Fourati
Contributed to the early design and implementation of a Planner-Executor-Critic multi-agent pipeline. Prepared and preprocessed evaluation datasets (AGIEval, LogiQA) and engineered role-specific prompts for each agent.
🔗 arXiv Contributor
🎓 Bachelor's Capstone Thesis  ·  Mediterranean Institute of Technology, 2024–2025
Ensemble Model Selection for LLMs: Does Picking the Best Model Improve Response Quality?
Yosr Fourati  ·  Supervised by Dr. Dorra Louati (MedTech) & Dr. Amine Barrak (Oakland University)
Introduced a dynamic routing framework that routes prompts to the best-suited LLM among four state-of-the-art models (DeepSeek V3, Claude-3.5 Sonnet, GPT-4o, Perplexity Sonar) across five task categories. The framework achieves win rates of 65–90% over individual models, with a top-1 accuracy of 74.6% using DeBERTaV3Small. Category-specific routing with Random Forest reaches 84.9% accuracy on coding tasks.
Outstanding Grade

// where I've worked

Industry Experience

Graduate Teaching Assistant
Oakland University · Rochester, MI, USA
Sep 2025 – Present
  • Assist undergraduate computing courses covering software systems, system integration, and web technologies.
  • Support lab sessions, handle grading, and provide one-on-one student mentoring.
  • Help students debug projects and understand core system architecture concepts.
Software Systems Web Technologies Mentoring
Full-Stack Engineer Intern
Bee Coders · Remote
Jun 2024 – Aug 2024
  • Designed and implemented end-to-end features for a multi-seller e-commerce platform using Java Spring Boot, Angular, and MySQL.
  • Completed 32 Jira tickets across core modules: authentication, product management, cart, checkout, order history, and role-based dashboards for Admin, Seller, and Customer.
  • Integrated Stripe payments, implementing full billing and payment flows with validation and error handling.
  • Built RESTful APIs, database schemas, and Angular frontend components from scratch.
Spring Boot 3 Angular 17 MySQL Stripe API JWT Auth REST APIs
Software Engineer Intern
STB Bank · Tunis, Tunisia
Jun 2023 – Aug 2023
  • Developed a Python script integrating with a legacy banking system to extract transaction and payment records via SQL queries, generating structured CSV/Excel reports for audit and operations end-to-end ownership of data extraction, cleaning, and formatted delivery to business stakeholders.
Python SQL Data Processing Legacy Systems
Software Engineer Intern
EPPM · Tunis, Tunisia
Jun 2022 – Aug 2022
  • Contributed to internal web application development as a second-year engineering student, turning feature requirements into working UI and backend updates.
  • Integrated frontend components with backend APIs and iterated on usability improvements based on feedback from internal users.
  • Gained hands-on exposure to real-world software workflows including version control, code review, and working within an existing codebase.
Web Development Backend Integration UI Development

// things I've built

Projects

🛒
⬡
ShopSphere

Full-stack multi-role e-commerce marketplace. Buyers browse and checkout, sellers manage listings and orders, admins monitor platform GMV and analytics. JWT auth, Stripe payments, 40+ REST endpoints, admin dashboard with Angular Signals.

Spring Boot 3 Angular 17 MySQL Stripe JWT Spring Security
🏠
Indeed Immobilier

Full-stack real estate agency platform for property listings and management. Includes property search, filtering, sale/rent workflows, and location metadata. Deployed to production with focus on performance and clean architecture.

Next.js TypeScript Tailwind CSS Node.js Prisma ORM PostgreSQL
🔗 indeedimmobilier.com
🎫
🏆 Best Senior Project Award
Event Management System

Comprehensive event platform with organizer dashboard and full attendee experience. Features QR-based check-in, event scheduling, ticket management, and post-event workflows. Awarded Best Student Project at MedTech (Senior year).

Award Winner
📱
🏆 Best Junior Project Award
KhademniApp

Mobile application connecting short-term employers and job seekers. Implemented user management, job discovery flows, and matching logic to streamline gig-style hiring. Awarded Best Student Project at MedTech (Junior year).

Award Winner Mobile

// tools of the trade

Skills

Languages

Java Python TypeScript JavaScript SQL C / C++

Frontend

Angular 17 React Next.js Tailwind CSS

Backend

Spring Boot Spring Security Node.js Express Prisma ORM REST APIs

Databases

MySQL PostgreSQL

Cloud & DevOps

AWS (EC2, S3, RDS, IAM) Docker Linux Git CI/CD

AI / ML

PyTorch TensorFlow scikit-learn LLM Routing NLP Pairwise Ranking

Tools & Workflow

Jira Postman Jest Shell Scripting OpenRouter API

// get in touch

Let's Connect

✉️

yosrfourati@oakland.edu

 💼

linkedin.com/in/yosrfourati

 ⬡

github.com/yosr-fourati