Hi, I'm Majd Zarai
AI Engineer/Co-founder of YottaNest
Core Technologies
Building the Future with AI
Majd Zarai
AI Engineer & Data Scientist
AI Engineer with a solid foundation in mathematics, ML, and deep learning. I specialize in LLM pipelines, RAG systems, graph intelligence, and multi-agent architectures. As Co-Founder of YottaNest, I focus on building scalable automation and high-performance intelligent systems.
Mission
Actively seeking opportunities to grow, learn, and contribute to impactful AI projects that shape the future.
Core Technologies
Location
Tunisia
Open to Remote
Education
ESPRIT University
Data Science & AI
Ready to collaborate?
Let's build something amazing together.
Featured Projects
A collection of AI and machine learning projects showcasing expertise in LLMs, computer vision, and intelligent systems.
Vigilant X – AI-Powered Crypto Fund Due Diligence
Automated crypto fund due diligence platform preventing fraud with AI-driven analysis and forecasting
YottaNest
AI-driven compliance and credit intelligence platform for banks and financial institutions.
GraphRAG Agentic Intelligence System
An agentic AI system using LangGraph, custom tools, and a Neo4j-powered GraphRAG pipeline.
Anomaly Detection System
A robust AI system for detecting rare events and hidden outliers in high-dimensional data.
Churn Prediction App
AI-powered customer retention tool with real-time predictions
Real-Time Parking Space Detection
Computer vision system for live parking spot monitoring
Recognition
2nd Place – IndabaX Tunisia 2025 Hackathon
Secured 2nd place out of 69 competing teams at IndabaX Tunisia 2025 hosted by SUP'COM. Developed advanced anomaly detection system using machine learning pipelines including Isolation Forest, Autoencoders, and statistical methods. Awarded 10,500 TND prize for outstanding performance in identifying rare events and outliers in high-dimensional data.
IndabaX Tunisia (SUP'COM)
Best Project of the Year – ESPRIT University 2025
Selected from over 200 competing projects for Vigilant X, a crypto fund due diligence platform addressing post-FTX fraud prevention. The platform automates fund analysis through PDF scanning, internet research, KYC/AML checks, and professional PPTX report generation. Uses Monte Carlo simulations for crypto forecasting with human-in-the-loop validation for audit-grade reliability.
ESPRIT University
Top Project Selection – Value Incubator
Recognized by Value Incubator for Vigilant X, a crypto fund due diligence platform preventing FTX-like fraud. Automates comprehensive fund analysis through PDF scanning, internet research, and KYC/AML compliance. Generates professional PPTX reports and uses Monte Carlo simulations for forecasting. Reduces manual processing time by 90% while handling 10,000+ daily queries.
Value Incubator
Zindi Global Leaderboard Recognition
Ranked #2028 worldwide on Zindi's global leaderboard as a Leading AI Practitioner. Achieved Gold medal (2/69) in IndabaX Tunisia 2025 Challenge 1 and Bronze medal (102/265, Top 30 Worldwide) in Landslide Detection Challenge, demonstrating competitive skill in machine learning and real-world problem solving.
Zindi Africa
Competitions:
IndabaX Tunisia 2025: Challenge 1
Rank: 2 / 69
Classification for Landslide Detection
Rank: 102 / 265 (Top 30 Worldwide)
Work Experience
- Architected advanced multi-agent AI platform automating KYC/AML workflows using Web, Graph, RAG, and Report Agents
- Integrated multiple LLMs (Llama 3.1, ChatGPT, DeepSeek, Kimik2, BERT) enabling adaptive reasoning and compliance automation
- Designed privacy-first microservices architecture ensuring secure on-premise data processing and regulatory alignment
- Achieved 90% reduction in manual processing time with improved accuracy in entity recognition and risk detection
AI Engineer Intern
Value Incubator- Architected Django application with advanced semantic chunking, RAG architecture, and automated fact verification for Vigilant X platform
- Implemented sophisticated PPTX reporting using LLMs integrated with real-time financial data APIs
- Deployed scalable microservices handling 10,000+ daily queries with 99.9% uptime
- Award: Best Project of the Year (ESPRIT 2025) — Selected from 200+ competing projects
- Developed state-of-the-art NLP models for real-time adverse media monitoring and risk classification
- Built intelligent web scraping agents with adaptive algorithms for continuous data enrichment
- Enhanced client decision-making by 40% through AI-driven insights integration into compliance workflows
- Designed end-to-end Tunisian Arabic voice assistant with integrated STT, NLU, and TTS pipelines
- Built comprehensive speech data pipelines including collection, preprocessing, and model fine-tuning
- Reduced response time by 60% while replacing traditional call center workflows with intelligent automation
Certifications
Building Transformer-Based NLP Applications
NVIDIA • 2025
Certification for completing NVIDIA's course on building NLP applications using Transformer-based architectures.
Applications of AI for Anomaly Detection
NVIDIA • 2025
Focused on developing AI systems for anomaly and fraud detection across real-world datasets.
Fundamentals of Deep Learning
NVIDIA • 2025
Comprehensive certification covering neural networks, backpropagation, training workflows, and deep learning fundamentals.
Bal des Projets – Participation Certificate
ESPRIT • 2025
Official participation certificate for the 12th edition of the Bal des Projets organized by ESPRIT.
Continuously expanding knowledge through industry-recognized certifications
Let's Connect
Have a project in mind or want to collaborate? I'd love to hear from you.
Available for Work
Open to full-time & freelance
Send a Message
Fill out the form below and I'll get back to you as soon as possible.