Hi, I'm

Daniel Diaz

Data Analyst · data quality, SAP and automation / Madrid, Spain

I make data trustworthy — and use AI where it earns its place: local LLM pipelines for duplicate detection, agents that automate the repetitive work, and products built with AI tooling.

SAP S/4HANA ↔ Sales Cloud sync accuracy from 45 % to 99.8 % · LLM-based duplicate detection in MDM · self-hosted AI assistant · Beacon TCG, built with AI agents

Resume →Projects LinkedIn ↗

SQL (T-SQL)✦Azure Synapse Analytics✦Python✦Power BI (dashboards)✦Alteryx Designer / Cloud✦Excel✦Data profiling✦Reconciliation✦Root cause analysis✦Quality KPIs✦SAP S/4HANA✦SAP Sales Cloud✦SAP GUI Scripting✦Reltio MDM✦Dun & Bradstreet✦SharePoint✦Power Automate✦LLM pipelines (local models via Ollama)✦Duplicate detection / record linkage✦Prompt & JSON output design✦AI-assisted development (coding agents)✦Master data✦Rule definition✦UAT / QA✦Data stewardship✦Control documentation✦GDPR✦TypeScript✦Next.js✦SQLite✦Docker✦Git✦SQL (T-SQL)✦Azure Synapse Analytics✦Python✦Power BI (dashboards)✦Alteryx Designer / Cloud✦Excel✦Data profiling✦Reconciliation✦Root cause analysis✦Quality KPIs✦SAP S/4HANA✦SAP Sales Cloud✦SAP GUI Scripting✦Reltio MDM✦Dun & Bradstreet✦SharePoint✦Power Automate✦LLM pipelines (local models via Ollama)✦Duplicate detection / record linkage✦Prompt & JSON output design✦AI-assisted development (coding agents)✦Master data✦Rule definition✦UAT / QA✦Data stewardship✦Control documentation✦GDPR✦TypeScript✦Next.js✦SQLite✦Docker✦Git✦

01About me

I'm a Data Quality Junior Specialist at Radisson Hotel Group: I make sure the data behind the group's B2B accounts (travel agencies and corporate profiles) holds up across SAP S/4HANA, SAP Sales Cloud and the MDM. I am the technical profile inside the business: I turn what Sales needs into rules, queries and controls, and argue them through with IT. I joined in 2024 as a trainee and moved to Data Quality four months later.

Before that I spent seven years running restaurants and a leisure park in London and Madrid, making daily calls on sales, costs and staffing from the numbers. I know what bad data costs because I have paid for it.

Outside work I build things: a self-hosted AI assistant where I try local models before trusting them at work, and Beacon TCG, a Pokémon card price tracker I run end to end with coding agents. And when there is no screen in front of me, bonsai: prune what does not belong, correct little and often, and accept that the good results take years to show.

Skills

Data

SQL (T-SQL)Azure Synapse AnalyticsPythonPower BI (dashboards)Alteryx Designer / CloudExcelData profilingReconciliationRoot cause analysisQuality KPIs

Enterprise systems

SAP S/4HANASAP Sales CloudSAP GUI ScriptingReltio MDMDun & BradstreetSharePointPower Automate

AI & LLMs

LLM pipelines (local models via Ollama)Duplicate detection / record linkagePrompt & JSON output designAI-assisted development (coding agents)

Data governance

Master dataRule definitionUAT / QAData stewardshipControl documentationGDPR

Other

TypeScriptNext.jsSQLiteDockerGit

02Featured projects

View all →

01Active

Beacon TCG

Pokémon card price tracker with the US and EU markets side by side, a daily data pipeline and a free collection manager.

TypeScriptNext.jsSQLiteTailwind CSSDocker

↗02Active

Self-hosted AI assistant

Personal Telegram assistant on a home PC with a local LLM, vision, Spanish/English voice, web search, automatic memory and service monitoring. Zero cost, no data leaves the house.

PythonasyncioOllamaqwen2.5moondream

↗03Active

SAP bulk-change automation (internal tool)

Python tool on SAP GUI Scripting for bulk changes in SAP S/4HANA from Excel, with dry-run, validation and a change log. Built at work, used across departments.

PythonSAP GUI Scriptingpywin32openpyxl

↗

03Latest posts

View all →

September 16, 2026·5 min read

Duplicate detection in master data with a local LLM: rules find the candidates, the model decides

How I built a duplicate-detection pipeline for customer accounts with Python and a local LLM via Ollama: blocking, pairwise classification with a reason, a human review queue, and why it runs on-prem.

data-qualitymdmllmpythonollama

↗02

September 15, 2026·4 min read

Bulk changes in SAP S/4HANA with Python and SAP GUI Scripting — no RPA licence, no ABAP

How I automated mass updates in SAP S/4HANA from Python using the SAP GUI Scripting API: dry-run, per-row validation, change log, and when this approach makes sense.

sappythonautomationdata-quality

↗

04Contact

A project, an offer or just a chat about data? Drop me a line and I will reply.

Also on LinkedIn ↗