Data fabric & Lakehouse Engineering

From the Warehouse Floor
to the Data Lakehouse.

Nearly 20 years running Retail Replenishment Operations, now building the Azure Fabric pipelines that give retail and distribution teams the same visibility over their data that I once had to build by hand on the floor.

Bronze
raw operational data
Silver
cleaned & modeled
Gold
decisions you can trust
About Onyx

An operator's eye, applied to data

Onyx Data Engineering is founded and run by Irina Cazacu.

I spent nearly two decades leading retail and replenishment operations — responsible for the flow of stock across stores and distribution centres, and for the reporting that told (or failed to tell) leadership what was actually happening on the ground. I know first-hand what it costs a business when data is scattered across disconnected systems, and what it's worth when it isn't.

That experience, combined with a background in Chemical Engineering, gave me a natural fit for Data Engineering: both disciplines are about designing reliable systems and controlling what flows through them. I'm now formalizing that into Microsoft Azure and Fabric expertise, and applying it specifically to the retail and distribution sector I know from the inside.

My focus is building lakehouse foundations, pipelines, and reporting layers that give operational teams the visibility I once had to piece together manually — done properly, on modern cloud infrastructure, the first time.

Background
Chemical Engineering
Operational experience
~20 years, retail & replenishment leadership
Cloud platform
Microsoft Azure & Fabric
Certifications
AZ-900, DP-900, DP-700 (in progress)
Focus sector
Retail & distribution
Based in
United Kingdom
The path here

Two disciplines, one habit of mind

Every stage of this path taught the same lesson from a different angle: a business runs on the quality of the information underneath it.

Foundation
Chemical engineering
Trained to think in systems, inputs, and process control — the same logic that now shapes how I design a data pipeline.
~20 years
Retail & replenishment operations leadership
Ran the operational side of retail and distribution — demand, stock flow, and the reporting gaps that come from data spread across too many disconnected systems.
2025 – 2026
Azure fundamentals
AZ-900 and DP-900 certified, plus a Firebrand Training Data Engineering Bootcamp, to formalize the cloud and data platform side of the work.
In progress — target Dec 2026 / Jan 2027
DP-700: Microsoft Fabric Data Engineer Associate
Deep specialization in Microsoft Fabric — lakehouses, pipelines, and the medallion architecture this site borrows its structure from.
How I can help

Engagements, structured like the data itself

Retail and distribution teams tend to need help at one of three depths. I've named them after the layers I actually build.

Bronze
Lakehouse foundation
Get scattered operational data into one governed home.
  • Microsoft Fabric / Lakehouse setup
  • Ingesting source systems (ERP, POS, inventory)
  • Establishing raw data landing zones
From £750 / project
Enquire
Silver
Pipelines & modeling
Turn raw data into clean, trustworthy, queryable tables.
  • PySpark / Fabric pipeline development
  • Data cleaning, transformation, and modeling
  • Automated refresh & orchestration
From £1,800 / project
Enquire
Gold
Strategic consulting
Data that's ready for the decisions leadership actually makes.
  • Replenishment & demand reporting layers
  • Dashboard-ready gold datasets
  • Advisory grounded in real ops experience
From £300 / day
Enquire

Prices are starting points — every engagement gets scoped to the size of your data and systems before a final quote. Ongoing support is also available as a monthly retainer from £600/month. Get in touch for a free 20-minute scoping call.

Selected work

Portfolio

Four projects built to show the full lakehouse lifecycle, from raw ingestion to a gold layer ready for reporting. Code and architecture notes are public on GitHub.

01
Lakehouse foundation
Standing up a Fabric lakehouse from scratch — bronze-layer ingestion, folder structure, and governance basics.
FabricLakehouse
02
Pipeline development
PySpark transformation pipelines that move raw data through cleaning and modeling into analysis-ready tables.
PySparkETL
03
Retail data modeling
A schema and modeling approach built around retail replenishment concepts — demand, stock position, and lead time.
Data modelingRetail
04
Reporting layer
Gold-layer datasets shaped for dashboards, closing the loop from raw ingestion to a decision-ready view.
Gold layerReporting
View the full repo on GitHub →
Credentials

Certifications

AZ-900
Azure Fundamentals — certified
DP-900
Azure Data Fundamentals — certified
DP-700
Fabric Data Engineer Associate — in progress