Class of ML

Build Real ML Systems.

Go beyond tutorials. Work on real Pakistani datasets, build real pipelines, ship real models. This is not a course — it is a builder program.

20–30 Seats · Cohort 1

The Program

What You'll Do

01

Real Datasets

Work with actual Pakistani data: air quality indices, court case records, agricultural yields. No toy datasets.

02

Build Pipeline

Design end-to-end ML pipelines from data collection to model deployment. Hand-draw your architecture, explain your decisions.

03

Ship Models

Deploy working models with real APIs. Your output is not a notebook — it is a production-ready system.

How You Get In

Your Selection Process

01

Application

Free · 10 min

Basic details + why you want to build ML systems. No GPA, no resume, no prior experience required.

02

Screening

Free · 30 min

Conceptual ML understanding. No code required. Diagnose training/validation gaps on Pakistani datasets.

03

Assessment

Rs. 1,000 · 48 hours

Predict whether tomorrow’s Lahore AQI will cross 200. Submit a clean repo, a hand-drawn pipeline diagram, and a 3-minute voice note explaining your approach.

The Output

What a Class of ML output looks like

48-Hour Assessment Build

Lahore AQI Predictor

train.py — output
Epoch 12/15 — loss: 0.0342 — val_loss: 0.0289 — val_accuracy: 0.891
Epoch 13/15 — loss: 0.0318 — val_loss: 0.0275 — val_accuracy: 0.897
Epoch 14/15 — loss: 0.0301 — val_loss: 0.0268 — val_accuracy: 0.903

Best model saved: lahore_aqi_v2.pt (val_acc: 0.903)

Hand-Drawn Pipeline

Data CollectionFeature EngineeringModel TrainingEvaluationAPI Deployment
GitHub RepositoryDeployed APIVoice Note Submitted

Ready to build real ML systems?

SELECTIONHACKATHONRESEARCHMENTORSHIPPUBLICATIONCONFERENCESELECTIONHACKATHONRESEARCHMENTORSHIPPUBLICATIONCONFERENCESELECTIONHACKATHONRESEARCHMENTORSHIPPUBLICATIONCONFERENCESELECTIONHACKATHONRESEARCHMENTORSHIPPUBLICATIONCONFERENCE