Building data pipelines that turn data into business value.
I'm Haroune Mohammedi
For the past seven years, I've been building large-scale data pipelines that businesses actually depend on. At Engie that means designing infrastructure that moves 50M+ rows a day of high-frequency energy data on Databricks and PySpark — and turning multi-week ETL jobs into sub-hour runs along the way. What I care about most is the bridge between technical decisions and business outcomes: pipelines that don't just move data, but deliver it reliably, on time, and in a form the business can actually act on.
I'm a Senior Data Engineer based in Paris with over 7 years of experience building data systems that actually work in production.
My current focus is large-scale pipeline engineering on Databricks and PySpark. At Engie, I work with high-frequency energy consumption data — 50M+ rows a day — designing pipelines that need to be fast, reliable, and maintainable. I've cut execution times from weeks to hours through legacy ETL migrations, and achieved 90%+ performance improvements through architectural redesign.
I operate across the full project lifecycle — from functional analysis and architecture design through to deployment and production support. I care about understanding domain challenges deeply and bridging the gap between technical implementation and real business needs.
Before that, I helped build an MLOps platform from scratch at BigMama Technology, and spent years deep in distributed systems with Scala, Akka, and Kafka. That foundation still shapes how I think about data architecture today.
Open to Senior Data Engineer and Data Architect opportunities where technical excellence and business impact go hand in hand.
I've wanted a HomeLab for years. Last year I finally built it, and
it's become my favourite project.
It's partly about privacy — I'm uncomfortable with how much of my
digital life runs on infrastructure I don't control. Self-hosting
is my answer. It's partly about independence — my data and
services exist because I built them, not because a company decided
to keep offering them.
But mostly, it's my lab. When I want to understand a technology, I
deploy it. Recently that meant running
local LLMs on my own
hardware. A 3-2-1 backup strategy keeps everything safe.
Infrastructure as code, reproducibility, resilience — the same
principles I care about professionally, applied at home.
A university project I'm still proud of — a video player
controlled entirely by hand gestures, built as a fully
distributed, reactive application.
Gesture detection, video control logic, and the UI each ran as
independent
Akka actors
communicating asynchronously. Akka Remote handled distribution
between components. The gesture recognition itself was a neural
network trained with Keras.
My first experience taking a trained model and integrating it into
a live application pipeline. At the time I didn't have the
vocabulary to call it MLOps — but the core challenge was exactly
what I'd spend years working on later at BigMama.
Open to Senior Data Engineer and Data Architect opportunities in Paris and beyond. Always happy to talk about data architecture, MLOps, or distributed systems.