The person behind the work
About
AI Engineer from Tunis, in AI since 2021: RAG, agents, edge vision. AI Team Lead at 1morething Studio. Shipped in 8 countries. Ready to relocate for the right team.
2021
In AI since · full-time from 2025
AR·FR·EN
Languages spoken
8
Countries shipped in
~50
Professionals trained
Identity
At a glance
Now
AI Team Lead @ 1MT
I own the AI roadmap across the 1MT venture portfolio and validate every product's AI architecture, while building hands-on: edge vision in Riyadh, Arabic RAG for FAO, an on-premise legal pipeline for a ministry. I mentor 2–4 engineers. Based in Tunis, open to relocating.
Education
Esprit Engineering
National Engineer Degree · Data Science · 2025
Certified
4× NVIDIA DLI
Deep Learning · NLP · Computer Vision · Anomaly Detection
Sport
Club Africain · Real Madrid
Tactics nerd. Also watches NBA; short but loves basketball.
Music
Mood-driven
Arabic pop · Algerian raï · whatever fits the task
Off-hours
Astronomy & rockets
Anything that leaves the atmosphere.
Services
What I get hired for
01
RAG in production
Retrieval that survives real users and real documents: chunking, hybrid search, re-ranking, evaluation. Including Arabic and dialect, which breaks most off-the-shelf pipelines.
02
LLM agents & MCP
Agents that call your systems instead of guessing, and the evaluation tasks that prove they work: I build agent-eval tasks for Anthropic's frontier models. Spring AI, MCP and A2A in Java, or Python where that fits your stack better.
03
Edge computer vision
NVIDIA Jetson, DeepStream, TensorRT. On-device inference for cases where the video cannot leave the building — retail floors, sites, facilities.
04
Data science
The unglamorous half that still pays: classification, forecasting, anomaly detection, and the data engineering underneath it.
05
MLOps & deployment
AWS SageMaker, Docker, Kubernetes. Getting a model out of a notebook and into something with logs, versions, and a rollback path.
06
Training your team
~50 professionals trained across 3 cohorts, in English, French, and Tunisian. The same works in-house: Python through Transformers, hands-on.
Principles
How I work
01
Ship first
The real feedback loop starts when a real user touches it, not in a doc.
02
Build for users
A demo that works 95% doesn't ship. Edge cases aren't edge cases in production.
03
Teach to sharpen
Running bootcamps in 3 languages forces clarity on what I actually know.
04
Stay curious
Spring AI, MCP, Arabic NLP: I chased them early. They keep becoming the work.
"Ship first, learn in production. The feedback loop only starts when something real is in someone's hands."