Data & AI — I build the pipelines, the models and the products on top
Nokia · Cleva · 3geeks
Where the work happened
Insurance software · ClevAI, its AI hub
Apprentice, Data & AI — ClevAI hub
Global telecom company
Data & AI Tools Specialist — internship, then apprenticeship
Staying on was a natural next step. I found a better fit at Cleva's AI hub.
Studio I co-founded · clients worldwide
Co-founder — studio & self-hosted production
I build data pipelines that pull many sources together, and the dashboards that make them readable — with an AI summary on top, so a conclusion no longer takes a dozen tabs.
Calling a hosted model is the easy part. I fine-tune open-source models, deploy them locally and integrate AI into real workflows — then show it working on the case at hand.
I conceptualise quickly and get a first working prototype into people's hands early, instead of reading documentation for months first. Clients decide faster when they can see a result.
Still an apprentice, still learning fast — and what I build now is the base of what I can bring to your team tomorrow.
Cleva Solutions — right now
Now · Cleva Solutions
AI engineering
Clients want to know what AI can do for them, and a slide will not tell them. I take existing open-source models, adapt them to a precise need, deploy them locally and put a working proof of concept in front of the client. It is where I work every day — and only the beginning.
From open model to client proof
OngoingOpen-source model
picked for the need
Adapt
fine-tuned to the use case
Deploy locally
on our own infrastructure
Client proof of concept
shown working, on their case
Nokia — 2025 · 2026
Internal platform · Nokia
Creator & lead developer
Information on a single topic was scattered across many internal tools, and every analysis meant reconciling them by hand. I built the pipeline that gathers it, links it together and presents it in one clean, live dashboard — with an AI summary on top.
Many sources → one clear view
7+ internal sources
scattered across tools
Collect
one automated pipeline
Correlate
one shared picture
Live dashboard + AI summary
read in one pass
3geeks — co-founded, built to be adapted inside companies
Product · also adapted inside Nokia
Multi-agent planning
A vague idea in a chat window never becomes a build plan. Prompt Hub turns a short brief into phased steps and copy-paste prompts, orchestrated by seven specialised agents over a dependency graph, with multiplayer editing. Built at 3geeks, then brought into Nokia.
Infrastructure
Co-founder & DevOps
Every 3geeks product runs on this. The studio's apps were scattered across Vercel with drifting environment variables; I consolidated them onto one self-hosted stack — Coolify, Traefik, Cloudflare Tunnel — with a straight path from git push to a live HTTPS domain.
Also shipped
A deterministic execution engine: a LangGraph workflow proposes, and a fail-closed risk manager alone approves. Four isolated modes, from fully mocked to live.
Rewrites a prompt for the same intent with fewer tokens, shows the carbon cost of each request and favours European models.
An AI voice agent that answers a restaurant's phone around the clock — takeaway, bookings, menu questions — then confirms by SMS and sends the order to the kitchen.
A regulated-sector SaaS guiding US couples through their own state's divorce paperwork. Multi-state compliance and data security from day one.
A fully local, OpenAI-compatible gateway in front of Ollama — personal tokens, live usage stats and model management for the whole studio.
The studio flagship: an intent-to-website generator where a written brief becomes a fully laid-out site.
In public
3geeks, still building
Prompt Hub — update
Prompt Hub — first look
Education
Mastère Data Engineering & AI — RNCP level 7
Data architecture, structural AI and cloud governance, alongside the Cleva apprenticeship.
Bachelor in Artificial Intelligence — licence grade
Final year as an apprentice at Nokia, closed by the best Bachelor project of the year.
Lycée Parc de Vilgenis, Massy
Electronics and embedded systems.
Toolbox
Certifications
IA & GenAI — Prompt Engineering N1
Liora · 2026
AWS Academy — Machine Learning Foundations
Amazon Web Services · 2026
AWS Academy — Cloud Architecting
Amazon Web Services · 2026
Python for Data Scientists
DataScientest · 2025
He is a particularly curious and autonomous young engineer with solid technical skills, especially in artificial intelligence. What sets him apart is his ability to take initiative, propose pragmatic solutions and go beyond what is asked of him.
He does not stop at understanding AI concepts — above all, he knows how to think about using them concretely to meet business needs and improve existing processes. He showed initiative and efficiency by building tools that genuinely helped tracking and decision-making.
An effective, autonomous apprentice with a solid command of AI. He thinks outside the box, proposes innovative solutions and turns an idea into a concrete result quickly. A promising talent — creative and impact-driven.
Elias proved his ability to listen and adapt time and again, delivering work that was consistently thorough and dependable.
Responsive team, a very clear process and clean execution. The new site smoothed out our acquisition.
They understood our business quickly and suggest choices that actually help. We saw real progress.
Roles, collaborations, or a question about something you saw here — my inbox is open.
