Fourth-year Software Engineering student @ INSAT · AWS Certified AI Practitioner
Full-stack engineer focused on applied AI/ML: document AI, LLM-powered products and ML for finance. Open to part-time and working-student roles, and to summer 2027 internships.
Who I Am
AI engineer in the making, building intelligent systems that actually matter.
I'm a fourth-year Software Engineering student at INSAT and an AWS Certified AI Practitioner. My work centres on applied AI: document AI with OCR, LLM-powered products and agents, computer vision, and machine learning for finance and risk. I care about taking models past the notebook and into production, with reliable pipelines, explainable decisions and solid tests. I thrive in fast-moving settings like hackathons and ML competitions, where hard constraints push me to build smarter solutions.
Selected Work
A full KYB compliance pipeline for French company documents (Kbis/RNE). It reads documents with OCR, sorts them using explicit rules, extracts fields with fuzzy matching and ICAO 9303 MRZ parsing, checks them against the official registry (data.gouv.fr), screens for sanctions, then scores each case as APPROVED, NEEDS_HUMAN_REVIEW or REJECTED. 311 automated tests.
Built during my Softcorp internship (fintech). Company code, so there is no public repository: OCR → rule-based classification → field extraction → registry cross-check → sanctions screening → decision scoring.
Credit Decision & Explainability Platform. Full-stack prototype using Python/FastAPI and Angular, leveraging Qdrant vector DB and AI agents for automated credit evaluation and decision explainability.
"Similarity Radar" for case explainability, automated risk scoring agents, and an AI-driven evaluation pipeline — built end-to-end in a competitive hackathon environment.
AI-powered real estate platform modeling the full property lifecycle. Features a personalized recommendation engine, KMeans + Gradient Boosting buyer profiling, CatBoost price estimation trained on Tunisian market data, an image intelligence pipeline (OpenCV, CLIP, NudeNet), and a LLaMA 3.3 70B conversational assistant via Groq.
Freight marketplace that matches truck drivers with clients to cut empty trips. It verifies documents with an LLM and takes online payments.
Production-grade insurance recommendation system combining XGBoost engine with GPT-4o Smart Assistant and automated MLOps monitoring suite.
Detects driver drowsiness and distraction in real time, with alerts and behavior scoring. Built during Hack For Good 4.0.
Secure banking platform: JWT with refresh-token rotation, multiple accounts and cards, paginated history, a UI for each role, and admin activation. Built during my BiatIT internship.
Full-stack academic platform for INSAT mixing three API paradigms: REST+JWT for mutations, GraphQL for on-demand document reads, and event-driven real-time via SSE, WebSockets & Webhooks. Includes a standalone NestJS Calendar Server with CRON-triggered signed webhooks for academic deadline alerts.
Online vinyl record store with product catalog, shopping cart, authentication, order management, and admin inventory control.
What I Work With
Background
Credentials
Under Pressure
Beyond Code
“AI is a Tool —Not a Replacement.”
How I Got Here
Started by actively participating in tech clubs, taking on leadership roles, and competing in intense hackathon environments to build real-world prototypes under pressure.
Focused on creating intuitive, responsive, and beautiful user interfaces using modern web design principles, frameworks like Angular, and UI libraries.
Mastered robust server-side technologies, developing secure APIs, managing complex databases, and delivering complete full-stack platforms with real users.
Currently deep-diving into AI and ML, exploring LLMs, vector databases, XGBoost, computer vision, and building intelligent data-driven solutions.