Only Few Seats Left

Module 1

Introduction to AI & Gen AI

·       Building an AI tool

·       Natural vs Artificial Intelligence, Brief history of AI, Weak vs Strong AI

·       AI vs Data Science vs Machine Learning vs Deep Learning

·       Data: Collection, Labelled vs Unlabeled, Structured vs Unstructured, Metadata

·       Overview of Machine Learning (Supervised, Unsupervised and Reinforcement Learning)

·       Overview of Deep Learning, Robotics, Computer Vision, Traditional ML, Generative AI

·       Generative AI: Introducing ChatGPT, Natural Language, Processing (NLP)

·       Large Language Models (LLMs): Training, N-Grams, RNNs, Transformers

·       Building LLMs: Prompt Engineering, Fine Tuning, RAG

·       Foundation Models vs Private Models

·       Inconsistency and Hallucination in Gen AI, Budgeting, Latency, Running out of Data

·       AI stack: Python, Working with APIs, Vector Databases, Hugging Face, LangChain

Module 2

Working with Python Libraries

• Introduction to Numerical Python (NumPy Library)

• Introduction and working with Pandas

• Series, DataFrames, working with various data types, Group-By operation

• Selecting a single column, important series methods

• Indexing & Sorting; loc & iloc with series; Inspecting DataFrames, filtering with conditional operators

• Regex: Anchors and Groupings, Range Expressions, Non-Greedy Matching, Substitutions.

Module 3

Natural Language Processing (NLP)

• Introduction, Supervised vs Unsupervised NLP

• Data preparation, Handling Stop words, Regular Expressions 

• Tokenization, Stemming, Lemmatization, N-grams

• Text tagging, Parts of Speech (POS) tagging, Name Entity Recognition (NER)

• Sentiment Analysis, Rule-based Sentiment Analysis, Pre-trained Transformers

• Numerical Representation of text, Bag of Words, TF-IDF

• Topic Modelling, Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA)

• Custom text classifier using various ML models     

Module 4

Large Language Models (LLMs)

• Understanding LLMs, General Purpose Models, Pre-training and Fine Tuning

• Deep- Learning recap, Transformer Architecture, Input Embeddings

• Multi-headed Attention, Feed-Forward Layer, Masked multihead Attention

• Understanding GPT, Open AI API, Generating Text, Customizing GPT output

• Key word text summarization, Coding a simple chatbot using LangChain in Python

• Hugging Face, Transformer Pipeline, Pre-trained tokenizers, Special tokens

• Q&A models : BERT architecture, Tokenizer, Embeddings, Calculating response

• Creating QA bot, BERT, RoBERTa, DistilBERT, GPT vs BERT vs XLNET, XLNET Embeddings and Fine Tuning

Module 5

LangChain

• LangChain Introduction, Tokens and Models, Setting up Environment, Open API key

• System, User and Assistant roles, Creating chatbot, Temperature, Max Tokens, Streaming

• LangChain Framework, ChatOpenAI, System, AI and Human messages

• Prompt Templates and Prompt Values, Few-shot chat message prompt templates

• String output parser, Comma-separated list output parser, Datetime output parser

• Piping a prompt, model and parser, Batching, Streaming, Runnable Sequence class

• Piping chains and Runnable Passthrough, Runnable Parallel, Runnable Lambda

• Retrieval Augmented Generation (RAG): Document Loading, Splitting and Embedding

• Document storing, retrieval and generation, loading with PyPDFLoader, Docx2txtLoader

• Splitting with character text splitter, markdown header text splitter

• Text embedding with OpenAI, Chroma vectorstore (Inspecting and Managing docs)

• Retrieval: Similarity search, Maximal Marginal Relevance (MMR) search, Vectorstore-backed retriever

• Generation: Stuffing documents and Generating a response 

Module 6

LangGraph

• States, nodes and Edges, First graph: Importing relevant classes, Building graph

• Conditional edges: Defining nodes, routing function, Building the graph

• Annotated construct and reducer functions, MessagesState, RemoveMessages

• Checkpointers and threads, Short-term memory with the InMemorySaver class

• The StateSnapshot class, Long-term memory with SQLite    

Module 7

Vector Databases

·       Database comparison: SQL, NoSQL and Vector, Understanding Vector databases,

·       Vector space: Introduction, Distance Metrics, Vector Embeddings

·       Vector database, comparison, Pinecone registration, walkthrough and creating index

·       Pinecone with Python: Connection, Pinecone Index, Upsetting data and using embedding

·       Vector database for Recommendation Engines, Biomedical research, Semantic search

Module 8

Speech Recognition Module

• Development and Evolution, Formants, Harmonics and Phonemes

• Sound and Sound waves: Fundamentals and properties

• Sample Rate, bit depth, bit rate, Audio signal processing for Machine Learning and AI

• Audio Features: Time-domain, Frequency-domain, time-frequency-domain, Fourier transform

• Acoustic and language modeling, Hidden Markov Models (HMMs)

• Traditional Neural Networks: CNNs, RNNs and LSTMs

• Advanced speech recognition systems: Transformers

• Building a Speech Recognition Model

• Audio file formats for speech recognition, Importing audio files in python

• Google Web Speech API, Evaluation metrics: WER and CER

• Dealing with background, Noise Creating spectrogram

• Whisper AI: Transformer-based speech-to-text, Transcribing multiple audio files

• Reversing the process: AI-powered text-to-speech

Module 9

LLM Engineering

• Hosting an LLM vs Using an API, Open-Source vs Closed-source, Tokens

• Pricing: Hosting an LLM vs Pay-by-Token, Initial Prompt Development

• Database Design and Schema Development, Activity Diagram

• OpenAI Playground, Optimizing Temperature, Top P for Different Use Cases

• Prompt Engineering for Software Lifecycle

• Streamlit: Introduction, Pros and Cons, Text Methods, Chat Elements, Session State

• Initializing an OpenAI Client, Implementing the Chat Functionality, Building the Setup Page

• Enhancing Chatbot Interaction with Session State, Feedback Functionality, Deployment

• Application Structure, Prompt Structure, Hallucinations, Prompt Injection

• Counting Tokens, Cost reduction and Scaling