✦ LIVE PORTFOLIO GUIDEASK ME ANYTHING ↘
AVAILABLE FOR NEW OPPORTUNITIES ✦

HELLO, WORLD. I'M

Rana
Romdhane.

AI / ML ENGINEER ✦ RESEARCHER ✦ BUILDER

I turn complex intelligence into human experiences. From vision-language research to real-world AI products, I build what comes next.

Ariana, Tunisia/Download CV
Pixel art avatar of Rana Romdhane
READY TO TALKAI-GENERATED VOICE

Ask about Rana's work and research. Answers may be imperfect; the voice is AI-generated. For direct contact, email Rana.

✦Rana's AI TwinExplore my work through conversation
OPEN ↗
✳

Hi, I'm Rana's AI twin.

Ask about projects, research results, tools, or the problems I love solving.

TEXT + VOICE
ENTER TO SEND · SHIFT+ENTER FOR NEW LINE

Voice questions and spoken replies use external speech services. Chat history stays in this tab.

01ABOUT ME

Curiosity, coded into impact.

Researcher. Engineer. Builder of useful AI.

I turn machine learning ideas into experiences people can understand and use—from vision-language research to production-ready products.

BASED IN ARIANA, TUNISIA · BUILDING FOR THE WORLD
EXPLORE THE THREE SIDES OF MY WORK01 / 03
Research / IN PRACTICE

Questions worth testing.

I study how vision-language models read complex human expression, then validate results with explainable AI.

96.81%compound emotion accuracy
0196.81%Emotion model accuracy
0296%RAG answer faithfulness
03200K+Financial records analyzed
042026Published VLM research
02THE JOURNEY

Experience that moves things forward.

Choose a chapter to see the work and its impact.

EXPERIENCE FILE / 010.997forecast R²

AI Engineer Intern (Final-Year Project)

SORETRAK· Kairouan, Tunisia·Feb 2026 – June 2026
  • 01

    Designed and delivered TransportAI, an intelligent public transport management platform integrating a React analytical dashboard, an ML prediction engine, an assisted planning system, and a breakdown management module.

  • 02

    Benchmarked Random Forest, XGBoost, Gradient Boosting, and Linear Regression for operational forecasting; selected Random Forest with R² = 0.997 to support data-driven planning.

  • 03

    Engineered a multilingual RAG assistant with LangChain, ChromaDB, and a locally deployed Ollama LLM, reaching 96% answer faithfulness on a 50-question evaluation set.

  • 04

    Architected a fully on-premise stack with React, FastAPI, Oracle 11g, and Ollama; built the responsive React front end consuming FastAPI REST APIs for data confidentiality and operational autonomy.

BUILT WITH
Predictive MLRAGReactFastAPIOn-premise
Explore related work
03SELECTED WORK

Proof of possibility.

A collection of research, deployed intelligence, and products built to solve real problems.

SHOWING 05 / 05 PROJECTS

Final-year project · Operational intelligence02 / 05

TransportAI

An on-premise public transport platform combining forecasting, assisted planning, breakdown management, and a multilingual AI guide.

Random ForestRAGLangChainChromaDB
0.997forecast R²
Machine learning · Cybersecurity03 / 05

Intelligent IDS

Real-time intrusion monitoring across eight attack families, with severity-based alerts and an ELK-connected dashboard.

Random ForestSVMNeural NetworksWebSocket
8attack families
Cross-platform product · React Native04 / 05

Tawa Food Kit App

A food kit delivery experience for iOS, Android, and web, from discovery and cart through ordering and location-based tracking.

React NativeExpo RouterTypeScriptNativeWind
3platforms
NLP · Productivity05 / 05

AI Task Prioritization

BERT-driven task prioritization and calendar automation in a GTD-inspired management application.

BERTNext.jsCalendar APIGTD
87%accuracy
04THE LIBRARY

Ideas worth sharing.

Open a research file or read the story behind a technical idea.

FILE 01 / 02
Research paper · COC 2026

Beyond Binary Emotions: A Multi-Architecture Vision-Language Approach for Compound Facial Expression Recognition

First-author research presented at the COC 2026 international conference in Senegal; compares CNN, transformer, and vision-language approaches to RAF-CE.

Read paper
FILE 02 / 02
Technical writing · June 2024

Unraveling the Web of Network Protocols

Published in Code Spectrum magazine by the IEEE Computer Society Chapter at ENICarthage.

Read article
05FOUNDATIONS

Built on curiosity.

From physics and chemistry fundamentals to software engineering and AI research.

01 / 02
ENGINEERING DEGREE

Engineering Degree in Computer Science (Diplôme National d’Ingénieur)

National Engineering School of Carthage (ENICarthage)

Charguia 2, Tunisia
02 / 02
PREPARATORY STUDIES

Preparatory Studies in Engineering (Physics–Chemistry Track)

Preparatory Institute for Engineering Studies of Monastir

Monastir, Tunisia
06THE TOOLKIT

Many tools. One mindset.

Choose a toolkit to see the languages, models, platforms, and systems behind the work.

SELECT A TOOLKIT ↘
TOOLKIT 01 / 06

Programming

Python
Java
SQL
JavaScript
TypeScript
CLICK ANOTHER CATEGORY TO EXPLORE✦ RANA'S TOOLKIT
ONE MORE QUESTION?

The conversation starts above.

Ask about the work, the research, or where I could help next.

Ask my AI twin
07NEXT CHAPTER

Let’s make something matter.

Have a challenge, collaboration, or interesting idea? I'd love to hear about it.

BASED IN ARIANA, TUNISIA ↗ OPEN TO THE WORLD
SEND A SIGNAL