FAQ
The questions people actually ask.
Split by who's asking. If yours isn't here, the assistant at the bottom right answers most things, and anything it can't it passes straight to me.
01 · FOR RECRUITERS & HIRING MANAGERS
Hiring questions.
What kind of role are you looking for?
Permanent Data Analytics & BI roles first — Power BI development, semantic modelling, SQL reporting. I also work on freelance projects and longer engagements, and I'm open to the right full-time role.
How much experience do you have?
4+ years in practice and 38+ projects delivered across GitHub, LinkedIn and Upwork, spanning BI reporting, forecasting and GenAI automation. Eight of those are written up as full case studies on this site.
Which of your certifications are exam-based?
Four. The Microsoft PL-300 (Power BI Data Analyst Associate) is a proctored vendor exam, and the DataCamp Data Engineer, Data Scientist and AI Engineer for Data Scientists certifications each involve assessments and carry their own credential IDs. The remaining ten are professional certificate programmes and skill tracks, labelled on the page with whoever issued them.
Do you work remotely?
Yes — remotely and end to end, planning calls around overlapping hours so async and live collaboration both work.
Are you available right now?
I'm open to new projects and roles. The quickest way to confirm current availability and timelines is a short message.
What's the fastest way to assess my work?
The case studies, then the live demos. Each case study names the decision that changed rather than the tool that was used — the ER analytics one is probably the clearest example, where a healthy-looking 35.3-minute average wait was hiding that only 40.68% of patients met the 30-minute target.
02 · FOR CLIENTS
Working together.
How do you price work?
It depends on scope. Most projects are quoted as a fixed price once the requirement is clear; ongoing support works better monthly or hourly. Send a short description of the problem and I'll come back with an honest estimate — including if I think it's smaller than you expect.
How long does a typical project take?
A focused dashboard on clean data is usually days. A warehouse plus a reporting layer is typically a few weeks. An agent or automation depends mostly on how many systems it has to touch.
How does a project actually run?
Five steps: diagnose what decision is blocked and what's in the data, model the warehouse and semantic layer, build in short reviewable cycles, validate at every layer, then hand over.
What happens at handover — am I locked in?
No. Everything is built to be handed over: transformation logic version-controlled in Git, documented layer by layer, with in-report glossaries where stakeholders need to understand how a metric is calculated. You own it afterwards.
Will you sign an NDA?
Happy to work under an NDA — several projects on this site are described without naming the client for exactly that reason. Client data stays in your environment wherever possible.
Do you work alone or with our team?
Both. I work solo end to end, which is the point — no handoffs between a data engineer, a BI developer and an AI specialist, because I cover all three. I also work alongside in-house teams where they already own part of the stack.
We're still running on spreadsheets. Is that a problem?
It's one of the most common jobs I do. The Financial Analysis case study replaced a monthly manual P&L rebuild with a pipeline that refreshes end to end on its own.
03 · TECHNICAL
Stack and capability.
What do you actually build in Power BI?
Semantic models on a star schema, DAX measures and time intelligence, row-level and object-level security, and the reports on top. Beyond building: publishing to Power BI Service workspaces and apps, deployment pipelines, gateways and scheduled refresh, incremental refresh on large models, composite models and aggregations, and performance tuning with DAX Studio, Tabular Editor and Performance Analyzer.
Do you build the data warehouse too, or only the reports?
Both, and that's deliberate. Most BI developers depend on a pipeline someone else owns. I build the medallion warehouse underneath — bronze, silver and gold layers in Microsoft Fabric, Azure Data Factory pipelines, dimensional models with SCD type 2 — so reporting is never limited by data quality outside my control.
Do you do machine learning and forecasting?
Yes — demand forecasting, churn and risk scoring, lead scoring and NLP sentiment, evaluated properly with MSE, RMSE and R² rather than guessed at. It's a distinct capability rather than a bolt-on: there's a dedicated Data Scientist / ML column on the skills page, and DataCamp Data Scientist and Machine Learning Engineer credentials behind it.
What does the "AI automation" side actually mean?
Production LLM systems, not demos: RAG pipelines (retrieval-augmented generation — retrieving your own documents so answers are grounded and citable), agentic workflows with tools and human approval before anything irreversible, structured output validated against a schema with an auto-repair pass, and cost-aware routing between frontier and local models. Ten of these are documented.
Can you work with our existing Power BI setup?
Yes. That's often the more useful engagement — auditing an existing semantic model, fixing definitions that disagree between departments, tuning refresh and query performance, and adding the governance that was skipped first time round.
Are the demos on this site real, or mock-ups?
They're working tools. The dashboard cross-filters, forecasts and detects anomalies on seeded synthetic data; the ETL explorer walks a real production run including a quality gate that quarantines a failing row; the agent workflow runs a full cycle with a schema check that fails and auto-repairs on retry. The agent demo is scripted and makes no API calls, and the page says so.
Still got a question?
Ask the assistant — it knows the projects, the stack and the certifications. Anything it can't answer goes straight to me with your question attached.