Aix-en-Provence, France Data platforms · Streaming · ML systems Rev. 2026.09

Natale Foata. Data & AI Platform Engineer.

The stream reacts to your cursor. Hover Tap fail. to take the consumer down.

Seven years building the systems that carry data and models to production, and making sure they stay correct, not just up.

Currently
Yubo, since 2025
Before
CMA CGM · Ornikar · Traxens
Since
2019
01 What I work on

The systems that carry data to production.

Six things I like working on. What they have in common: a crash gets noticed, a wrong result does not. I care about the second one.

  1. 1.1

    Streaming and event pipelines

    Events moving between systems in real time. My job is that they arrive once, in order, and can be replayed when something breaks. One of the topic I enjoy most.

    With
    Kafka · Kafka Streams · Redis · Solace
    Where
    Traxens · Yubo
  2. 1.2

    Data reliability

    A job that finished is not a job that was right. I write the tests, checks and replays that catch the failures nobody sees: half-read inputs, duplicates, stale values.

    With
    dbt tests · dead-letter queues · backfills · reprocessing
    Where
    Everywhere
  3. 1.3

    ML platforms and model deployment

    Taking a model out of a notebook and putting it in front of real traffic, then knowing when it starts to drift. Classic ML first, embeddings and LLMs more recently.

    With
    MLOps · model serving · embeddings · feature pipelines
    Where
    Everywhere
  4. 1.4

    Databases and distributed stores

    Relational since day one. NoSQL is more recent: how these stores replicate, how they fail, and what a migration does to a cluster that is also serving users.

    With
    MongoDB · Elasticsearch · Redis · PostgreSQL
    Where
    Yubo · Marty
  5. 1.5

    AI engineering

    Models on images, prompt-injection detection, real-time signals feeding a recommender. On Valyra, a sentence without an official source does not get published.

    With
    LLM pipelines · vision models · prompt safety · evaluation
    Where
    Yubo · Valyra
  6. 1.6

    Infrastructure and cost

    Infrastructure as code, CI/CD, and knowing what each pipeline costs. Nobody notices this layer until the invoice arrives.

    With
    GCP · Kubernetes · Terraform · Docker · GitLab CI
    Where
    Every role
02 Where I have been

Seven years, only one goal: production.

Firefighters, consulting, three data platforms, a year on my own. Every move was about doing something I had not done yet. Scrub the timeline: each step says what I did there and what I took from it.

Yubo Data Platform Engineer

The most demanding environment so far.

The real-time data platform of a high-volume social app: streaming, storage, recommendation and the infrastructure behind inference. The richest and most technical environment I have worked in, and the one where mistakes cost the most. That is exactly why I chose it.

  • Streaming & event pipelines
  • Data infrastructure
  • Data reliability & analytics

Today

Deliberately narrow. Depth on the systems that carry data and models to production, rather than breadth across whatever is new. And, since Marty, a habit of asking who the system is for before touching it.

03 Who I am

The rest of the day.

I grew up in the south of France and never really left it. I studied mathematics, theoretical first, then applied to AI and signal processing, with a year in a Marseille fire station in the middle.

I have trained since I was fifteen: lifting, boxing, running. CrossFit and Olympic weightlifting are the current chapter, with first competitions planned for 2027 (mainly to have fun with friends).

I have played World of Warcraft since 2009. Woltk is the one I loved, which is why I play Classic: what I like in an MMO is the people, and Classic still has them. Less time than before, so casual now, on a shaman, healing. StarCraft II, Overwatch, Minecraft and Space Marine 2 take turns with it.

Off screen: Martin and Tolkien on the fantasy side, Kingdom and Baki on the manga side.

Lives in
Aix-en-Provence, France
Studied
Mathematics, then AI & signal processing · 2014 – 2020
Training
CrossFit · Olympic weightlifting · first competitions in 2027
Playing
World of Warcraft Classic (shaman, healer) · StarCraft II · Overwatch · Minecraft · Space Marine 2
Reading
G. R. R. Martin · Tolkien · Kingdom · Baki
Speaks
French · English
04 Writing
  1. Sept 2026 What a year building Marty taught me about engineering I built a product, finished it, and discovered that finishing was never the hard part. What a year on my own changed in how I work. Building
All writing ↗
05 Contact

Let's talk systems.

Streaming, data platforms, ML in production, or how you keep a system honest at scale. I enjoy these conversations for their own sake.

Email
foatanatale@gmail.com
LinkedIn
linkedin.com/in/natale-foata
GitHub
github.com/Natale-F
Feed
/rss.xml
Off call
CrossFit & weightlifting · World of Warcraft · StarCraft II · Overwatch · Space Marine 2