Ayesha Khan · Karachi, Pakistan · open to data roles

From business to data analytics.

The fourth industrial revolution didn't just change the tools we work with — it changed who gets to use them. Data isn't an IT skill anymore; it's a working language, and anyone with a real problem to solve can learn it. I studied business. I spent years marketing data & AI to companies. Now I'm building the skills to do the work myself — and sharing every step here.

Explore projects Read my story

BBA · Bahria University
2.5 yrs marketing a data & AI consultancy
11 hands-on data projects

Analytics · 3D bar chart

How I work

Tools change. The loop doesn't.

Every project on this page follows the same discipline: start with a business question, let the data argue, and end with a decision someone can actually act on.

01 · Data

It never arrives clean. Nobody's does.

Nulls, duplicates, a price of $0 and a date from the year 2099. The first job is to understand what the data can and cannot say — and to write down every fix as the business assumption it really is.

customer_id	plan	tenure	churned
C-1041	monthly	1	yes
C-1042	NULL	14	no
C-1041	monthly	1	yes
C-1043	annual	-3	no
C-1044	Monthly	2	yes

Found: 1 duplicate, 1 null, 1 impossible value, 1 inconsistent label. Logged, fixed, documented.

02 · Insight

Churn isn't spread evenly. It has an address.

Customers on monthly plans leave in their first two months at three times the rate of everyone else. That's not a statistic — it's an onboarding problem wearing a churn costume.

M7+

Monthly-plan churn, months 1–2: 3.1× the baseline. Annual plans: flat.

03 · Decision

Move the retention budget to week one.

Instead of discounting everyone at month six, trigger onboarding nudges and a check-in call in the first fourteen days for monthly-plan customers — and measure it against a holdout group.

ActionWeek-1 onboarding sequence

Target segmentMonthly plan · tenure < 2

Projected impact−12% churn

How we'll knowA/B test · 8 weeks

Approved for pilot

This loop auto-plays. Click any step to explore it.

About

I didn't leave business. I gave it better tools.

BBA, MarketingBahria University · 2024

Now: data analyticsPython · SQL · Power BI

Next: AI & engineeringPyTorch · Airflow · dbt

I graduated with a BBA in Marketing from Bahria University in 2024, and spent two and a half years as a Marketing Specialist at traxccel, a data & AI consultancy. My job was to explain how data transforms companies. Somewhere along the way, explaining stopped being enough — I wanted to build.

So I started from the ground up: Python and Pandas for the mechanics, SQL for the questions, statistics for the honesty, Power BI for the story. Then machine learning, deep learning and generative AI for the predictions — and data engineering for the plumbing that makes any of it real.

The advantage of coming in from business is simple: I already know what a dashboard is for. I've sat in the meetings where the numbers had to change a decision. Now I'm learning to be the person who gets those numbers right.

Business fluency

I know the question.

Margins, funnels, retention, positioning — and how a decision actually gets made in a room.

Technical depth

I'm building the answer.

Python, SQL, statistics, ML and the pipelines that turn a one-off analysis into a system.

The overlap

The translation layer — where data becomes a decision.

Most projects fail in the gap between “the model says” and “so we should”. That gap is exactly where I want to work.

The journey so far

Business → Data → AI → Engineering. Click a stage.

Skills

A toolkit, not a badge wall.

Honest levels. “Confident” means I've shipped it in a project; “Building” means I'm using it weekly; “Exploring” means I'm early and say so.

Projects

Learning in public.

Personal projects built on public datasets — Kaggle, UCI, GroupLens and open government data. Each one starts with a business question and ends with a decision. Flip the switch to read the full story.

Playground

Data in motion.

Small interactive pieces I built to understand what I was learning — a dashboard, a theorem, a query, a network and a pipeline. Every one of them runs live in your browser.

Revenue overview

A Power BI-style report, rebuilt in plain SVG. Switch regions.

Dashboard

Monthly revenue

Revenue by category

Live — refreshed just nowSuperstore-style sample data

Central Limit Theorem, live

Sample means become normal — whatever the source.

Statistics

Samples0

Mean of means–

Std. error–

This is why a 30-customer survey can still tell you something true about all customers — with a confidence interval you can defend.

A question, in SQL

Which states get the most late deliveries — and how do they rate us?

SQL

FROMJOINWHEREGROUP BYORDER BYLIMIT

state	late_orders	avg_review

Typing query…

How a model makes a prediction

Three facts about a customer go in. A churn probability comes out.

Machine learning

Churn risk — move a slider to predict

Tenure (months) 12 Monthly charge ($) 70 Contract Monthly

Illustrative weights — the point is the mechanics, not the prediction.

From sources to decisions

An ETL pipeline feeding a warehouse feeding a dashboard — the shape of every data platform I'm learning to build.

Data engineering

ExtractPull from CRM, ERP and web events0 rows

TransformValidate, dedupe, type, enrich0 rows

LoadIdempotent upserts into the star schema0 rows

ServeOne definition of every metric, in the dashboard0 refreshes

Experience & education

Where I've been.

The business years are the foundation, not a detour. Everything on this page is built on them.

Work

Education & certifications

Let's talk

Let's turn data into decisions.

I'm looking for data analyst and junior data scientist roles or internships — remote or in Karachi. If you have a messy dataset and a real question, I'd love to hear about it.