Artificial Intelligence Engineer

Building AI systems
from research to production.

I’m Joseph Elias Al Khoury, an AI Engineer focused on multimodal machine learning, LLM pipelines, semantic search, medical imaging, and production-ready AI products. I work across text, images, time series, and tabular data to turn complex signals into useful systems.

15+ GitHub repositories
4 Core data modalities
3 Languages spoken
AI → Prod Research, backend, deployment
Portrait of Joseph Elias Al Khoury
Based in France AI Engineer · ML · LLM Systems
Open to impact Healthcare · SaaS · Applied AI

About

AI Engineer with experience in data science, machine learning, and Python development. I design and deploy models for heterogeneous data, including images, text, time series, and tabular datasets, and build reproducible pipelines using tools such as Docker, Git, FastAPI, and vector databases.

My work sits at the intersection of applied research and production systems: from radiomics and survival prediction in medical imaging to RAG assistants, semantic retrieval, and intelligent SaaS features.

LLMs & RAG Multimodal ML Medical Imaging FastAPI Docker Vector Search
Current focus AI products & applied ML

Building LLM pipelines, semantic search workflows, and scalable AI systems that move beyond notebooks into usable products.

Research edge Healthcare AI

Working on survival prediction, radiomic feature engineering, and clinically oriented multimodal modeling.

Engineering mindset Reproducibility first

Strong emphasis on clean pipelines, benchmarking, deployment, and making AI systems robust enough for real environments.

Experience

Applied AI experience spanning startup products, medical research, and production pipelines.

AI Engineer & Co-founder · SISSI France · 2026 – Present

Built LLM pipelines for note structuring, summarization, extraction, and prioritization. Implemented semantic search with embeddings and pgvector, designed ingestion-to-insight pipelines, and integrated AI modules into a React, Django REST, and Supabase architecture.

Data Scientist in Medical Imaging · Institut Curie (Inserm U1288) Orsay, France · 2024 – 2025

Developed a multimodal survival prediction model for lung cancer using MRI, PET, and clinical data on 180+ patients. Defined and validated new radiomic features integrated into LIFEx, with a scientific manuscript in preparation.

AI Intern · Institut Curie (Inserm U1288) Orsay, France · 2024

Rebuilt a radiomics extraction pipeline on brain MRI, optimized it through multi-CPU parallelization, reduced runtime by 40%, containerized the workflow with Docker, and developed predictive models for genomic mutations.

Featured Projects

AI Portfolio Agent

Live ↗

Autonomous AI agent that analyzes, summarizes, and enhances developer portfolios using LLMs. Designed to evaluate projects, generate insights, and suggest improvements for stronger personal branding.

LLMs AI Agent Automation Python Prompt Engineering

Polished

Live ↗

Full-stack AI assistant integrating a Chrome extension (MV3) with a FastAPI backend using Gemini, providing contextual text rewriting, translation, and persistent knowledge capture directly in the browser, with structured note management and export capabilities.

Chrome Extension (MV3) FastAPI Gemini API Full-Stack AI Knowledge Capture

LegalChatbot

Intelligent Lebanese legal assistant built for Arabic question answering using a RAG architecture combining SBERT, Gemini, FastAPI, and React.

RAG SBERT Gemini FastAPI React

MediNetFusion

Multimodal time-series classification project for early type 2 diabetes prediction using 77 biomarkers over 12 months with models including CNN, LSTM, and XGBoost.

CNN LSTM XGBoost Time Series Healthcare AI

SISSI AI Assistant

AI-powered executive assistant SaaS centered on note structuring, semantic retrieval, and insight generation through modern LLM workflows.

LLMs Embeddings pgvector Django REST Supabase

Hospital Flow Optimization

Optimization work on emergency department logistics using real-world hospital data, scheduling, prediction, multi-agent systems, and IoT-driven analysis.

Optimization Q-Learning Multi-Agent Healthcare Ops

LLM Picker Extension

Utility project focused on improving day-to-day AI workflow ergonomics through a practical extension-oriented interface.

Python Extensions AI Workflow

Gym Membership Manager

Software project demonstrating practical backend and data handling skills through a management-focused application.

Python Management App Backend Logic

Skills

Core engineering stack across machine learning, MLOps, data systems, and software development.

AI / ML

PyTorch, TensorFlow NLP, LLMs, prompt engineering RAG, embeddings, semantic search Supervised & unsupervised learning Multimodal modeling

MLOps & Deployment

FastAPI, Streamlit, Gradio Docker, Git, Linux shell MLflow Elasticsearch, pgvector Reproducible pipelines

Data & Research

NumPy, Pandas, scikit-learn Feature engineering & benchmarking Spark, Power BI Medical imaging & radiomics Optimization algorithms

Programming

Python, Java, C SystemVerilog, MATLAB, R, SQL HTML, CSS GitHub, GitLab, VS Code SPARQL, UML / SysML

Education

Master’s AI École Centrale de Lille & University of Lille · 2023 – 2025

Graduated with High Honors, with advanced work in artificial intelligence applied to healthcare systems.

Bachelor’s in Biomedical Engineering Lebanese International University · 2020 – 2023

High Honors. Built a strong foundation in electronics, microcontrollers, signal processing, robotics, and medical systems engineering.

Languages

Arabic — Native English — C1 French — C1

Career Direction

Looking to contribute as an AI Engineer, Applied AI / LLM Engineer, or AI-focused Data Scientist in teams building useful, scalable systems.

Let’s build something serious.

Interested in AI engineering, LLM systems, multimodal learning, healthcare AI, or production-ready ML? Reach out through any of the channels below.

Quick Snapshot

Specialties Applied AI

LLM systems, semantic retrieval, multimodal prediction, radiomics, and deployable ML backends.

Style Engineer mindset

Focused on performance, structure, reproducibility, and practical impact rather than flashy demos.

Environment Cross-domain

Comfortable in startup contexts, research labs, and interdisciplinary teams solving hard real-world problems.