$ whoami
A data science student who believes the most dangerous thing you can do with data is present it without understanding it.
I build systems that turn noise into signal — currently in the part of the story where the protagonist is still training.
$ cat /etc/philosophy
Most people treat data science as a toolbox.
I treat it as a lens.
The goal isn't to run models.
The goal is to ask the question no one thought to ask —
then answer it rigorously.
Automation is not laziness. It is respect for your own attention.
A dashboard is not a deliverable. It is a conversation.
Code that works is the floor, not the ceiling.
$ ls -la ./stack/
| Layer | Tools |
|---|---|
| Language | Python · C++ · SQL |
| Data | Pandas · NumPy · Scikit-learn |
| Viz | Matplotlib · Seaborn · Power BI |
| Infra | Git · Linux · Jupyter · VS Code |
| Automation | n8n · Playwright · Selenium |
| Learning | MLflow · FastAPI · Streamlit [ in progress ] |
$ cat ./now.log
[STATUS] Building in public — slowly, deliberately, without shortcuts.
[FOCUS] End-to-end ML pipelines · EDA that actually tells a story
[REGION] Pakistan · Local data. Global methods.
[NOTE] Most repos are private — being rebuilt to a higher standard.
They'll return when they deserve to be seen.
$ grep -r "signal" ./values/
Precision over speed. A model that explains itself is worth more than one that doesn't.
Context is everything. The same number means different things in different datasets. I try to never forget that.
Ship when it's ready. Not when it's done — when it's ready. There's a difference.
$ ping ./reach-me
╔──────────────────────────────────────────────────────────╗
│ │
│ "Not yet finished. Never actually done." │
│ │
╚──────────────────────────────────────────────────────────╝
Pakistan · 2025 → ∞