[Ep.1-2] The Technical Interview That Turned Into a Data Analytics Masterclass

It started with one interview — and turned into my personal analytics lab.

I’m building Wealth Tracker to make managing money easier, and FKE to make learning French more useful — including for DELF and TCF preparation. Both come from the same idea: reduce the friction between knowing and doing. Want to follow the build? See what I’m working on →

🍋 Section 1: The Interview That Sparked It

It started with a single email — the kind that makes your stomach flip in both directions.

I had landed a technical interview with a national restaurant brand. The role was exactly what I’d been looking for: part marketing strategist, part data analyst, part behavior decoder. The kind of hybrid role I’d quietly hoped existed — and now, I had a shot.

They gave me a case study: three hours to analyze the effectiveness of a marketing campaign using raw data, SQL, and Power BI. I’d just wrapped up a five-month data science bootcamp and had been binge-practicing campaign workflows for weeks. I opened my Notion, rolled up my sleeves — and dove in. And I loved it.

I didn’t get the job.

But the rejection lit a spark. It left me with a question:

What if I kept going?

That case study had revealed something I didn’t know I was looking for — a new way of thinking. A world where marketing wasn’t just instinct or design, but also structure. Where campaigns could be tested, measured, and refined — not just imagined.

I couldn’t let it go. So I didn’t.

Before I knew it, I’d built out a whole ecosystem: behavioral segments, campaign logic, messy data quirks, and just enough chaos to feel real. I drank more matcha than I care to admit.

I generated the data, cleaned the tables, mapped the relationships, and built dashboards in Power BI as if I were running the campaign myself.

If life gave me lemons, I wasn’t just making lemonade — I was naming the flavors, segmenting the drinkers, and tracking their retention curve.

This blog walks you through how I did it — from messy SQL to polished insights, and what I learned along the way.

Table of Contents

1. The Interview That Sparked It

2. I Didn’t Get the Job. So I Built the Job.

3. Designing My Own Dataset (Gemini + Python)

4. PHASE 1: Cleaning Data in SQL

5. PHASE 2: Modeling in Power BI

6. PHASE 3: Creating Relationships

7. PHASE 4: Funnel Analysis

8. PHASE 5: The Story — Visualizing the Insights 

---

🍋 Section 2: I Didn’t Get the Job. So I Built the Job.

At first, I told myself this would be a quick project.

That same day of the interview, I came home with the case paper still in my bag — a challenge I couldn’t let go of. I opened my laptop, launched ChatGPT and Gemini, typed in a few prompts, and started building something new. A dataset inspired by the interview — but expanded into a full business scenario: Customers. Transactions. Campaigns. Locations. Even weather patterns.

I planned to build 2–3 dashboards. Just a portfolio piece. A tidy case study to prove I could work with real-world data. But the deeper I went, the more questions I wanted to answer.

What would I need if I were running this campaign from start to finish?

So I designed for mess. For nuance. For behavior.

I built the data to breathe — not just sit in tidy rows.

• Customers with incomplete contact info

• Campaigns with overlapping dates

• Promotions that worked for some segments but not others

  
  

And I realized: this wasn’t just a practice exercise.

This was the job.

Not the title or the offer — but the mindset.

The habit of asking better questions.

The discipline to build systems that could answer them.

.

.

.

So I kept going.

Over the next two weeks, I built:

• 5 interrelated tables — customers, transactions, products, campaigns, weather

• Dozens of SQL cleaning queries

• A full Power BI model — with calculated measures for revenue, AOV, and campaign ROI

• 10 dashboards — each telling the story from a different lens: executive, marketing, product, analytics

It was thrilling at first. Then frustrating. Then thrilling again.

I redesigned my dashboards. Then redesigned them again. And again.

I forgot what day it was. I caffeinated like a startup founder.

There were moments I felt exhausted.

After all, there was no guarantee any of this would pay off.

But I didn’t quit — because what I was building felt true to how I see the world:

as a series of stories, systems, and subtle patterns waiting to be uncovered.

I didn’t get the job.

But in the process of trying — I built the kind of work I want to keep doing.

---

Next: [Ep3]: Designing My Own Dataset (Gemini + Python), we shift gears.

This is where I show you how I did it —from dataset generation, structured the workflow, and designed the logic behind it all.

I’m building Wealth Tracker to make managing money easier, and FKE to make learning French more useful — including for DELF and TCF preparation. Both come from the same idea: reduce the friction between knowing and doing. Want to follow the build? See what I’m working on →