Module 1: NLP Fundamentals
Introduction to NLP – Understanding how machines process human language.
End-to-End NLP Pipeline – From raw text to model prediction.
Text Preprocessing – Tokenization, stopwords, stemming & lemmatization.
Text Representation – Converting text into numerical vectors.
Word2Vec (CBOW & Skip-gram) – Deep understanding of word embeddings.
Text Classification – Spam detection, sentiment analysis & categorization.
POS Tagging – Identifying grammatical roles of words.
Duplicate Question Pairs – Detecting semantically similar questions.
Module 2: Neural Networks & Transformers
Artificial Neural Networks – Brain-inspired learning systems.
Activation Functions – Sigmoid, Tanh, ReLU & Leaky ReLU.
Loss Functions – MSE, MAE, Huber, Hinge, BCE & CCE.
Backpropagation – Core learning algorithm of deep networks.
Learning Rate, Epochs & Batch Size – Training fundamentals.
History of Large Language Models (LLMs).
Encoder–Decoder Architecture – Sequence-to-sequence models.
Attention Mechanism – Learning what matters most.
Bahdanau vs Luong Attention – Comparative study.
Transformers – Backbone of modern NLP & GenAI.
Self Attention & Scaled Dot Product Attention.
Multi-Head Attention – Parallel attention learning.
Positional Encoding – Injecting sequence order.
Layer Normalization – Stable training.
Masked Self Attention & Cross Attention.
Transformer Decoder & Inference Process.
Module 3: Large Language Models (LLMs)
How LLMs Work – High-level internal overview.
LLM Parameters – Temperature, Top-P, Stop Sequences & more.
LLM Hallucination – Causes & mitigation techniques.
Module 4: Prompt Engineering
Prompt Engineering – Designing effective prompts from beginner to pro.
Module 5: Word Embeddings & Similarity
Word Embeddings – Dense vector representations.
Word2Vec Intuition – Semantic relationships.
CBOW & Skip-gram – Internal working.
One-Hot Encoding & Bag of Words.
N-grams – Contextual learning.
TF-IDF – Importance-weighted embeddings.
Word vs Sentence Embeddings.
Cosine Similarity & Distance Metrics.
Module 6: Generative AI & LangChain
GenAI Roadmap – End-to-end learning path.
Introduction to LangChain.
LangChain Components – Models, prompts, chains & tools.
Structured Output & Output Parsers.
Document Loaders & Text Splitters.
Vector Stores & Retrievers.
Tool Calling & AI Agents.
Ollama – Running powerful local LLMs.
Module 7: APIs, Deployment & MLOps
API Fundamentals & Architecture.
FastAPI – High-performance Python APIs.
HTTP Methods – GET, POST, PUT, DELETE.
Pydantic – Data validation & schemas.
Serving ML Models with FastAPI.
Docker for Machine Learning.
FastAPI + Docker Deployment.
Deploying AI APIs on AWS.
Projects & Practical Implementation
Real-World AI & NLP Projects – Each student will work on industry-relevant problems using real datasets.
Minimum 5+ Projects per Student – Projects will be built after completing the modules to ensure strong practical understanding.
NLP-Based Projects – Text classification, duplicate question detection, sentiment analysis, and POS tagging based systems.
LLM & GenAI Projects – Prompt engineering use cases, LLM-based applications, and AI assistants.
LangChain Projects – Building document-based Q&A systems, AI agents, and retrieval-augmented generation (RAG) pipelines.
Deployment-Focused Projects – Serving ML models using FastAPI, Docker, and deploying APIs on cloud platforms.
EMS-Based Project Submission – All projects will be submitted, tracked, and evaluated through the EMS system.
Performance Review & Feedback – Each project will be reviewed and feedback will be provided for improvement.