TSH

Data Science

Build models that predict, classify, and discover patterns in data.

Advanced9 to 14 monthsanalyze
Your Progress0%

Overview

Data scientists build machine learning models that predict outcomes, classify data, and find hidden patterns. The field sits at the intersection of statistics, programming, and domain expertise.

Who is this for?

  • Students starting from zero who want structure
  • Self-taught learners who need a clear path
  • Career switchers ready to commit consistently

What can you build?

  • Predictive models
  • Recommendation systems
  • Image classifiers
  • NLP pipelines
  • Fraud detection systems

Jobs you can get

Data ScientistML EngineerAI ResearcherNLP EngineerComputer Vision Engineer

Salary range: $70,000 to $160,000/year

Roadmap

Follow each stage in order. Mark stages complete as you finish them.

  1. Stage 1Complete first

    Prerequisite: Data Analysis

    • Python + Pandas
    • SQL
    • Statistics basics
    • Data visualization
  2. Stage 23 to 4 weeks

    Statistics & Math

    • Probability distributions
    • Linear algebra basics
    • Calculus concepts (gradients)
    • Hypothesis testing
  3. Stage 36 to 8 weeks

    Machine Learning

    • Supervised learning: regression, classification
    • Unsupervised: clustering
    • Model evaluation metrics
    • Feature engineering
    • Scikit-learn
  4. Stage 44 to 5 weeks

    Deep Learning Basics

    • Neural networks fundamentals
    • TensorFlow or PyTorch
    • CNNs for image tasks
    • Intro to NLP
    • Transfer learning
  5. Stage 52 to 3 weeks

    MLOps & Deployment

    • Saving and loading models
    • FastAPI ML endpoints
    • Model monitoring
    • Experiment tracking (MLflow)

Resources

Projects

beginner

#01

Titanic Survival Predictor

Classic Kaggle competition. Build a classifier with feature engineering.

PythonPandasScikit-learn

intermediate

#01

House Price Prediction

Regression model predicting property prices with feature analysis and cross-validation.

Scikit-learnPandasEDA

advanced

#01

Image Classifier

CNN classifier for 10-class image recognition using PyTorch.

PyTorchCNNGPU Training
#02

Deployed ML API

Train a model, save it, and serve predictions via FastAPI endpoint.

FastAPIscikit-learnDocker

AI Guide

What AI helps with

  • Explaining ML algorithms in plain terms
  • Debugging model training code
  • Suggesting feature engineering ideas
  • Explaining statistical concepts

What AI cannot replace

  • Replacing mathematical understanding, you must grasp why models work
  • Validating results without your own analysis

Recommended Tools

Sample Prompts

Explain gradient descent as if I understand basic algebra but not calculus. Include why it works and what can go wrong with learning rate.
My Random Forest has 97% training accuracy but only 61% test accuracy. Explain what is happening and give me 5 concrete steps to fix it.

Career Path

Data ScientistML EngineerAI ResearcherNLP EngineerComputer Vision Engineer

$70,000 to $160,000/year

Kaggle competitions are your portfolio. Even finishing in the top 40% shows employers you can apply ML to real problems. Publish your notebooks.