gammaticatech

Python for AI & ML

Learning Format

Live Online / Classroom

Total training duration

120 hrs

Syllabus

12 weeks

Certification

Yes

Python for AI & ML

Python for AI & ML is widely used because of its simplicity and powerful ecosystem of libraries. It provides tools like NumPy, Pandas, Scikit-learn, TensorFlow, and PyTorch for building intelligent systems. Python helps in data preprocessing, model training, evaluation, and deployment with ease. Its flexibility allows developers to experiment with algorithms for classification, prediction, and deep learning.

Syllabus Summary

Python recap for ML (10 hours)

Objectives: Bring everyone to a consistent Python/ML environment; basic numerical computing.

  • Topics: Python refresher (functions, OOP basics), virtual environments, Jupyter/Colab, pip, Git basics.

  • Libraries: NumPy (arrays, broadcasting), Pandas (DataFrame ops, indexing, missing data).

  • Lab (4 h): Data cleaning workflow — load CSV, explore, impute, scale, feature engineering.

  • Assignment (2 h): Clean & document a messy ML dataset (deliver cleaned CSV + notebook).

  • Deliverable: Cleaned dataset + brief data quality report.

ML intro & data splitting (10 hours)

Objectives: ML pipeline, supervised vs unsupervised, holdout strategies.

  • Topics: ML workflow, features/labels, train/validation/test splits, cross-validation, pipelines.

  • Lab (4 h): House price prediction (simple linear regression baseline using scikit-learn).

  • Assignment (2 h): Build baseline model, report metrics and error analysis.

  • Deliverable: Notebook with train/val/test split, baseline model, metrics.

Regression models (10 hours)

Objectives: Linear & non-linear regression techniques and regularization.

  • Topics: Linear Regression, Polynomial features, Ridge, Lasso, feature selection, bias-variance tradeoff.

  • Lab (4 h): Salary prediction project (feature engineering + Ridge/Lasso).

  • Assignment (2 h): Compare models, cross-validate, submit short report.

  • Deliverable: Notebook + comparison plot/table.

Classification models (10 hours)

Objectives: Classification algorithms and evaluation.

  • Topics: Logistic Regression, k-NN, Decision Trees, Random Forests, ROC/AUC, confusion matrix, class imbalance strategies (resampling, class weights).

  • Lab (4 h): Iris classifier + multiclass handling.

  • Mock Interview 1 (2 h): Short technical + practical coding question.

  • Assignment (2 h): Build & evaluate a classifier with hyperparameter tuning.

  • Deliverable: Notebook + model evaluation summary.

Clustering & unsupervised methods (10 hours)

Objectives: Unsupervised learning methods and when to use them.

  • Topics: K-means, Hierarchical clustering, DBSCAN, PCA for dimensionality reduction.

  • Lab (4 h): Customer segmentation (cluster analysis + interpretation).

  • Assignment (2 h): Create clusters, visualize, and profile segments.

  • Deliverable: Notebook with cluster visualizations and business insights.

Model evaluation & deployment basics (10 hours)

Objectives: Robust evaluation and basic model deployment concepts.

  • Topics: Cross-validation in depth, stratified sampling, metrics for various tasks, model selection, overfitting remedies. Intro to model serialization (pickle, joblib) and serving options (Flask/Streamlit basics).

  • Lab (4 h): Fraud detection dataset — work on imbalanced classification, precision/recall tradeoffs.

  • Assignment (2 h): Build best performing model and export it; write inference demo.

  • Deliverable: Exported model file + inference notebook/web demo.

Neural networks basics (10 hours)

Objectives: Introduce deep learning concepts and build an ANN.

  • Topics: Perceptron, multilayer perceptron, activation functions, forward/backprop intuition, loss functions, optimizers (SGD/Adam), regularization (dropout, batch norm).

  • Framework focus: TensorFlow (Keras Sequential) or PyTorch (basic Module).

  • Lab (4 h): Build a simple ANN for a tabular classification/regression task.

  • Assignment (2 h): Implement network training loop, log metrics, and plot learning curves.

  • Deliverable: Training notebook + saved model.

CNN basics (10 hours)

Objectives: Convolutional Neural Networks for image tasks.

  • Topics: Convolutions, pooling, architectures (LeNet/VGG concepts), data augmentation, training tips.

  • Lab (4 h): MNIST digit classifier (or small CIFAR subset) with augmentation and callbacks.

  • Mock Interview 2 (2 h): Model design + debugging challenge.

  • Assignment (2 h): Improve accuracy via augmentation/regularization; produce confusion matrix.

  • Deliverable: Notebook + model metrics.

NLP preprocessing & classic NLP pipeline (10 hours)

Objectives: Text preprocessing and classical approaches for text classification.

  • Topics: Tokenization, stemming/lemmatization (NLTK & spaCy), stopwords, TF-IDF, bag-of-words, text normalization.

  • Lab (4 h): News classification (preprocess, vectorize, train a classifier).

  • Assignment (2 h): End-to-end text pipeline and model evaluation.

  • Deliverable: Notebook + preprocessing pipeline code.

Word embeddings & sequence models (10 hours)

Objectives: Dense representations and basic sequence models for NLP.

  • Topics: Word2Vec / GloVe / pretrained embeddings, Embedding layers, simple RNN/LSTM/GRU intuition, classification with embeddings; sentiment analysis pipeline.

  • Lab (4 h): Sentiment analysis using pretrained embeddings or embedding layer.

  • Assignment (2 h): Build & compare embedding strategies (pretrained vs learned).

  • Deliverable: Notebook + comparative analysis.

Transfer learning & advanced topics (10 hours)

Objectives: Leverage pretrained models, introduction to fine-tuning and lightweight deployment.

  • Topics: Transfer learning for vision (feature extraction vs fine-tuning), model checkpoints, saving/loading TF/PyTorch models, lightweight inference (TensorFlow Lite mention), ethics & bias in ML.

  • Lab (4 h): Mini chatbot (intent classification + simple rule-based responses) OR image transfer learning (fine-tune a pretrained model on small dataset).

  • Assignment (2 h): Small project showing transfer learning benefits.

  • Deliverable: Notebook + fine-tuned model.

Capstone & interview prep (10 hours)

Objectives: Integrate everything in a final project; final mock interview.

  • Topics: Project presentations, code review, reproducibility, model cards, deployment demo (basic).

  • Capstone Projects (choose one or two):

    • Sentiment Analysis (end-to-end: data → model → deployment demo)

    • Image Classifier (transfer learning, deployment demo)

    • Sales Prediction (time series + regression pipeline)

  • Mock Interview 3 (2 h): Full technical + system design + HR style questions.

  • Deliverable: Capstone project repo, presentation, README + demo.

Edit Template

Course Summary

Eligibility

Tech & Non-Tech Working professional, Freshers, Graduate from any domain.

Live Doubt Solving

Get your queries solved with daily dedicated doubts solving sessions.

Instructor

Experts and trainer for top-tech companies.

Certification

10+ ISO Globally recognized certified

Mode of Learning

100% Live Learning with experienced instructors and hands-on sessions.

Real time projects

Get practical experience with real-world projects for a career in analytics.

Certification

Scroll to Top