I see this question every single week in Discord servers,
Telegram groups, and college WhatsApp chats:
"Bhai kahan se start karu AI/ML?"
[Translation: "Where do I start with AI/ML?"]
And I see the same bad advice every time:
"Just start with Andrew Ng"
"No, do Krish Naik first"
"Just learn Python and figure it out"
Nobody gives a complete, opinionated, ordered answer.
This is mine.
PREREQUISITES (before Month 1)
- Basic programming understanding (any language)
- A laptop with 8GB RAM minimum
- 2 hours per day commitment
- That's it. No Math degree needed. No CS background needed.
MONTH 1: Python Foundations
Topics:
- Variables, data types, control flow
- Functions, recursion, lambda
- Object-Oriented Programming (classes, inheritance)
- File I/O, exception handling
- List comprehensions, generators
- NumPy and Pandas basics
Project to build:
A data analysis script that reads a CSV,
cleans it, and outputs summary statistics.
MONTH 2: Mathematics for ML
Topics:
- Vectors and matrices (Linear Algebra)
- Derivatives and gradients (Calculus)
- Probability distributions
- Bayes theorem
- Mean, variance, standard deviation, correlation (Statistics)
Don't overdo this. 3–4 weeks is enough.
You'll understand more math as you apply it
in algorithms — it clicks better that way.
MONTH 3–4: Machine Learning
Topics:
- Linear & Logistic Regression
- Decision Trees & Random Forests
- Support Vector Machines
- K-Nearest Neighbors
- K-Means Clustering
- PCA (Dimensionality Reduction)
- Model evaluation: accuracy, precision, recall, F1, ROC-AUC
- Scikit-learn end to end
Projects:
- House price prediction (regression)
- Email spam classifier (classification)
- Customer segmentation (clustering)
MONTH 5: Deep Learning
Topics:
- Perceptrons and multilayer neural networks
- Activation functions, loss functions
- Backpropagation (understand it, don't just use it)
- CNNs for image classification
- RNNs and LSTMs for sequential data
- PyTorch framework (learn this, not TensorFlow)
- Batch normalization, dropout, regularization
Projects:
- Image classifier (CIFAR-10)
- Sentiment analysis with LSTM
MONTH 6: NLP & Computer Vision
Topics:
- Tokenization, stemming, lemmatization
- TF-IDF, Word2Vec, GloVe
- Transformer architecture (read the original paper)
- BERT and fine-tuning pre-trained models
- Object detection (YOLO)
- Image segmentation (U-Net)
- HuggingFace library
MONTH 7+: Generative AI
Topics:
- Large Language Models (GPT architecture)
- Prompt engineering techniques
- Retrieval Augmented Generation (RAG)
- Vector databases (Pinecone, ChromaDB)
- Langchain framework
- Building and deploying AI apps
This is the most employable skill in tech right now.
Don't skip it.
RESOURCES
Free:
- YouTube: CodeWithHarry (Python), Krish Naik (ML/DL/GenAI), CampusX (Data Science), Apna College (Python/DSA)
- Coursera: Andrew Ng ML Specialization (audit for free)
- Fast.ai (Deep Learning)
- Kaggle Learn (hands-on practice)
Organized Bundle (paid, ₹399):
If you don't want to spend weeks organizing all of
the above yourself → bundle.scrunchcreate.com
All the above educators, structured in this exact
order, one-time payment, lifetime access.
FINAL ADVICE
Don't collect resources. Build things.
For every 10 hours of watching → spend 10 hours coding.
The people who break into AI jobs aren't the ones
who watched the most tutorials.
They're the ones who built the most projects.
Now close this article and open your code editor.
Top comments (0)