Available · full remote since 2015

Senior Engineer · Python · Backend · Data

From ingestion to interface, I build the whole data chain.

25 years of Python in prod: PagesJaunes, Airbus, EDF, Deutsche Telekom. Systems that take the load and run on their own for years. The homelab hosting this site is the latest illustration.

My personal projects, in my lab, at home:
From the raw Linky frame to per-appliance consumption disaggregation.
From 16th-century handwritten records to an interactive genealogy graph.
Raw data turned into information that is useful to the user.
Each time, a base problem, what worked, what broke and the questions still open.

01
·VA
Data pipeline · Time-series

LinkyStat

GitHub

Turn a raw meter feed into usable data.

→ Many years in production. The data foundation for what came next.

→
02
demo data
· kWh avg / day
IoT · Home automation

Home Assistant

Measure individual appliances, not just the total.

→ Local IoT stack. And a blind spot that raises the real question.

→
03
·VA
ML · Time-series

Linkya · NILM

GitHub

Detect an appliance from the aggregate Linky signal alone: NILM.

→ Appliance detections published to Home Assistant. The consumption blind spot is closing.

→
04
CPU
RAM
Disk
Rx ·Tx ·
Backend · Infra · Prod

Self-hosted infra

Run my projects simply and reliably, at home.

→ Raspberry Pi, reverse proxy, auto TLS and CI/CD.

→
05
Applied AI · Graph model

Genealogy

Trace the family tree from genealogy and archive sites. Decipher handwritten records few people can still read.

→ A graph of 1,200+ people across 17 generations, from the 16th century to today, AI-assisted.

→
About

End-to-end engineer, from sensor to model to deployment.

I work at the intersection of hardware, data and software. I like problems where you have to understand the physics as much as the code, and ship something simple and reliable that lasts.

Roland Saïkali