Open to opportunities

Mahdi Messai

@mahdimessai

Message

AI engineer specializing in scalable ML systems and data-driven products.

Tunisia

Message

What I'm looking for

I seek roles where I can build and deploy production ML systems end-to-end, collaborate on data engineering and MLOps, and deliver measurable business impact in a growth-oriented team.

I am an AI engineer focused on building scalable machine-learning systems and data products that deliver measurable business impact. I combine hands-on ML model development with backend engineering to deploy production-grade solutions.

At my current role I architected and deployed a multi-agent AI system with retrieval-augmented generation (RAG) that improved performance by 20% on football-knowledge benchmarks, and built a FastAPI backend with load-balanced endpoints for high-throughput usage.

My background includes applied analytics work such as a Marketing Mix Modeling project, developing a VB.NET reporting solution that reduced report generation time by 60%, and research forecasting natural-gas consumption using Bayesian Additive Vector Autoregressive Trees in R.

I enjoy end-to-end projects—from data ingestion and orchestration with Airflow and dbt to model training with PyTorch and XGBoost and deployment using Docker and AWS—delivering production dashboards and interactive web apps that turn insights into action.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Calgra Group Holdings Inc.

Feb 2025 - Present (8 months)

Architected and deployed a multi-agent RAG-based AI system that outperformed base models by 20% on football-knowledge benchmarks; built a FastAPI backend with load-balanced endpoints and designed a real-time analytics pipeline using Airflow and dbt.

RAG fastAPI Airflow DBT Python Vector Databases Load Balancing

Data Analysis Intern

Mass Analytics

Jun 2024 - Sep 2024 (3 months)

Applied statistical-learning techniques for Marketing Mix Modeling to quantify media impact; developed a VB.NET reporting solution that reduced client-report generation time by 60% and delivered model-validation POCs to external clients.

Statistical Learning VB.NET R Python Validation

Research Analyst

Independent Research

Forecasted G20 natural-gas consumption using Bayesian Additive Vector Autoregressive Trees and produced reproducible analyses with R, tidyverse, and ggplot2 to inform energy consumption insights.

R tidyverse Ggplot2 Data Visualization

Education

Degrees, certifications, and relevant coursework

Tunis Business School

Bachelor of Science, Analytics

Grade: 3.6/4.0

Activities and societies: Relevant coursework: Time Series Analysis, Data Mining, Object-Oriented Programming, Web Services; student projects include DariPrice real-estate pricing model and KhedmAlytic job-market dashboard.

Completed a Bachelor of Science in Analytics with coursework in Time Series Analysis, Data Mining, Object-Oriented Programming, and Web Services; graduated with a 3.6/4.0 GPA.