03 — About

About.

Engineer from Télécom Saint-Étienne, trained in France and Japan. Independent through ELGOTE.

AI & Data engineer from Télécom Saint-Étienne.

I build complete data chains: from raw collection to a reliable decision you can put in front of the business. API ingestion, modelling, quality tests, orchestration, delivery.

What putting these systems in production taught me is that the hard part is almost never the technique. It is reliability. Wrong data that looks right costs more than missing data: it spreads, it ends up in a board meeting, and nobody knows where it came from. Hence my obsession with tests at every boundary, replayable pipelines and business definitions written down in black and white.

Same principle with AI. I use LLMs in production, but never alone on a decision: a model always produces an answer, including when it is wrong. My method is a deterministic funnel first — rules, exact matches, reliable sources — and the LLM last, on what remains, with human validation where it matters. Framed like that, it is a remarkable tool.

My ground: Python, SQL/PostgreSQL, dbt, DuckDB, Dagster, Docker, medallion architectures, Supabase, n8n.

Currently independent on data & AI engagements, and curious about everything around agents in production: security, observability, governance.

  1. Oct 2026 —

    ELGOTE

    AI & Data Engineer (independent)

    Data engineering and applied AI engagements: pipelines, product analytics, business-process automation, LLMs in production.

  2. Apr — Sep 2026

    Meet My Mama

    Data & AI Engineer (internship)

    Took over the company's whole data scope after the data team left: BI, warehouse, automations, AI.

    • Native analytics module inside the internal ERP — materialized views, scheduled refresh, drill-down — progressively replacing the external BI tool.
    • Warehouse migration: every dashboard and export rebranched to the new production database without downtime, with a rollback net at each step.
    • Accounting mirror: daily ingestion from the invoicing software's API, reconciled to the cent with the books.
    • Referential quality: CRM deduplication and venue cleanup through a deterministic cascade, an LLM only on the residue, corroboration required before any write.
  3. 2025 — 2026

    豊橋技術科学大学 · Toyohashi University of Technology

    Exchange — robotic perception, 3D vision

    Research exchange in Japan. EKF-SLAM implemented from scratch for the Robotic Perception class.

  4. 2023 — 2026

    Télécom Saint-Étienne

    Diplôme d'ingénieur — computer science, networks & telecom

    Engineering degree, diploma conferred March 2027. Top 15%.

  5. 2021 — 2023

    Lycée Joffre, Montpellier

    CPGE — advanced mathematics & physics

    Two years of classes préparatoires at competition intensity.

Earlier: Technical Sales Advisor (B2B), Kiloutou (Summer 2025) · R&D Intern, video surveillance, STIM (Summer 2024)

French & Arabic — native · English — C1 · Spanish — B2 · Japanese — B1, actively learning