Applied Data Science with AI/ML & Gen AI – 150 hrs.
Level 1 – 60 Hrs
Level 2 – 60 Hrs
Level 3 – 30 Hrs
Program Structure
- Machine Learning – 60 hrs.
- Supervised & Unsupervised Learning
- Regression
- Classification
- Model Improvement
- Deep Learning – 30 hrs.
- Tensorflow
- ANN (Artificial Neural Networks)
- CNN (Convolutional Neural Networks)
- OpenCV
- Natural Language Processing – 30 hrs.
- Tokenization
- Named Entity Recognition (NER)
- NLTK
- Gen AI & Agentic AI – 30 hrs.
- Transformers
- RAG (Retrieval-Augmented Generation)
- Prompt Engineering
Prerequisite:
- Python Programming
- Data Analysis using Pandas
Program Objectives
The program emphasizes end-to-end development
workflows, model evaluation, and deployment using
modern frameworks like Streamlit and Hugging Face,
preparing learners for successful careers in data science
and AI engineering.
Program Outcomes
By the end of this program, learners will be able to:
- Build and evaluate Machine Learning models for regression, classification, and clustering to solve real-world business problems.
- Develop and deploy Deep Learning models using CNNs and frameworks like TensorFlow, PyTorch, and Keras.
- Apply Natural Language Processing techniques including tokenization, stemming, embeddings, and text classification using ML and RNNs.
- Understand and implement Generative AI concepts such as GANs, VAEs, Transformers, and Prompt Engineering.
- Design and deploy AI applications using tools like Streamlit, Hugging Face, LangChain, and OpenAI.
Project Stream:
• Machine Learning and Deep Learning Deploy
using Streamlit or Huggingface
Experiential Project Based Learning
Tools / Platform:
• Google Colab /Jupyter Notebook
• OpenAI, Gemini, Copilot
• HuggingFace, Langchain, CrewAI
| Machine Learning (60 hrs.) | ||
|---|---|---|
| Understand what is Machine Learning | Regression | Logistic regression |
| Supervised machine learning | Simple linear regression | Naïve Bayed Classification |
| Unsupervised machine learning | Multiple linear regression | Decision tress and its types |
| Train test split the data | Performance measure for regression | K Nearest Neighbor Classification |
| ML Workflow for project implementation | MSE, R-Squared, MAE, SSE | Performance Measure for Classification |
| Classification | Various types of classification | Accuracy, Recall, Precision, F1measure |
| Sorting Data Frames | Importing csv files | Importing Excel Files |
| Deep Learning (30 hrs.) | ||
| What is Deep Learning | Hidden Layers | Building project based on CNN |
| Deep Learning Methods | Activation Function | Tensorflow, pytorch, Keras |
| Deep Learning Application | Forward and Backward propagation | Batch Normalization, dropout |
| Artificial Neural Network | Deep Learning Libraries | Performance measure for ANN |
| CNN architecture | CNN for computer vision | Object detection |
| Computer vision basics | OpenCV | Working with Images |
| Natural Language Processing (30 hrs.) | ||
|---|---|---|
| Natural Language processing | NLP Applications | Regular Expression |
| Tokenization, stopwords | Stemming, Lemmatization | Word Embeddings |
| NLTK | POS Tagging NER | bag of words, TF-IDF, unigrams, bigrams |
| RNN, RNN architecture | Bidirectional LSTM – Encoders and Decoders | Text classification using ML |
| Generative AI & Agentic AI (30 hrs.) | ||
| Introduction to Gen AI | Rule-based vs neural generation | Generative Adversarial Network |
| Variable Auto Encoder | Transformers | Application of Generative AI, Ethics |
| Sentence embeddings and similarity | Encoding long text documents | Visualizing embeddings with tools |
| Prompt Engineering | Zero – shot and few – shot prompts | Chain-of-thought prompting style |
| Retrieval-Augmented Generation (RAG) | Agentic Architectures & Autonomous Workflows | Multi-step Reasoning |
| Project | ||
| Apply statistical methods to make decision in various business problems, including bank, stock market | Apply regression to predict future flight price | Apply classification to classify customer churn |
| Streamlit Framework | Deployment using Hugging Face | Model and agent deployment |
