Module 1: Introduction to Deep Learning and AI (4 Classes)
Class 1: Introduction to AI and Machine Learning
● Overview of AI, ML, and DL
● Key Concepts and Terminologies
● Historical Context and Evolution
● Key Concepts:
○ Generative AI
○ LLM
○ Vector Database
○ Hugging Face
○ LangChain
● Importance of Kaggle profile.
○ Kaggle Competition
● The job of DL, LLM, Generative AI
[Link]
Class 2: Basics of Neural Networks
● Artificial Neurons
● Activation Functions
○ Linear, Sigmoid, Softmax, Tanh
○ ReLu, Leaky ReLu,
● Dying Relu Problem
● ANN Architecture
● Forward and Backward Propagation
● Training Neural Networks with Python
Class 3: Deep Learning Frameworks and Tools
● Introduction to Popular Frameworks
○ Keras
○ TensorFlow
○ PyTorch
● Setting up the Environment
● Basic Operations
● Model Creation with Python
Class 4: Training Deep Learning Models
● Data Import, Preparation, and Preprocessing
● Loss Functions and Optimization Algorithms
○ Gradient Descent Optimizer
○ Variants of Gradient Descents (Momentum, Nesterov Momentum,
AdaGrad, RMSProp, Adam and Nadam)
● Gradient Problems (Vanishing & Exploding)
● Key Concepts of-
○ Overfitting, Underfitting, and Bestfitting
○ Regularization Techniques
Module 2: Computer Vision (8 classes)
Class 5: Introduction to Computer Vision
● Overview of Computer Vision Tasks
● Image data Handling
● Data Augmentation
Class 6: Convolutional Neural Networks (CNNs)
● CNN architecture and components
● Convolution and pooling layers
● Fully connected layer
Class 7: Advanced CNN Architectures
● Popular CNN models (LeNet, AlexNet, VGG, ResNet, Inception)
[Link]
● Transfer learning
● Fine-tuning
Class 8: Object Detection and Localization
● Techniques (R-CNN, Fast R-CNN, Faster R-CNN, YOLO)
● Implementation and applications
Class 9: Semantic Segmentation and Image Segmentation
● Techniques (U-Net, Fully Convolutional Networks)
● Practical examples and use cases
● Implementation with Python
Class 10: Generative Adversarial Networks (GANs) in Computer Vision
● Introduction to GANs
● Architecture
● Training of GANs with Python
Class 11: Applications for GANs in Computer Vision
● Variants of GANs (DCGAN, CycleGAN, StyleGAN) & Image generation
and transformation
● Style transfer and super-resolution
● Training stability and challenges
● Implementation with Python
Class 12: Computer Vision Projects
● Implementing a real-world project
● Best practice and troubleshooting (Modular Code)
● Project Name: Automatic Dhaka traffic detection using the YOLO model.
Module 3: Natural Language Processing (NLP) (7 classes)
Class 13: Introduction to NLP
● Overview of NLP tasks
● Text preprocessing techniques
● Regex
● Implementation with Python
Class 14: Word Embeddings and Representations
● Tf-idf, Word2Vec, GloVe, FastText
● Contextual embeddings (ELMo, BERT)
● Implementation with Python
[Link]
Class 15: Recurrent Neural Networks (RNNs) and Variants
● Basic RNN architecture
● Long Short-Term Memory (LSTM)
● Gated Recurrent Unit (GRU)
● Implementation with Python
Class 16: Seq2Seq Modeling, Attention Mechanisms, and Contextual Embeddings
Attention Mechanisms and Transformers
● Sequence-to-Sequence Models for Neural Machine Translation (NMT)
● Attention mechanism
● Deep Dive into Contextual Embeddings
● Implementation with Python
Class 17: Advanced Transformer Models & Extended Contextual Embeddings
● Transformers in depth
○ Input Embeddings
○ Positional Encodings
○ Self-Attention, Multi-Head Attention
○ Encoder
○ Decoder
○ Output Layer
● Transformer Variations: Encoder only, Decoder only, Encoder-Decoder,
and their applications
● Extended Contextual Embedding Techniques with Transformer Model
● Evaluate NLP models
Class 18: Transformer Model Pretraining, Fine-Tuning, and GPT Decoding
● Pretraining Transformer Models
● Fine-tuning techniques
● GPT Decoding Strategies (Greedy, Beam Search, Sampling)
● Implementation with Python
Class 19: End-to-End NLP Project
● Implementing a real-world project
● Best practice and troubleshooting (Modular Code)
● Project Name: Word Spelling Correction
Module 4: Generative AI (7 classes)
Class 20: Introduction to Generative AI
● Overview of generative models
● Instruction Tuning (Basic & Advanced Prompt Engineering)
● Evaluation of LLMs (Metrics and Benchmarks)
● Applications and cases
[Link]
Class 21: Multimodality - Variational Autoencoders (VAEs) and Multimodal LLMs
● Understanding Multimodal Inputs (Text, Image)
● VAE Architecture demonstrates Multimodal Data
● Integrating Multiple Modalities into LLMs
● Applications for Multimodal LLM-powered chat assistant
Class 22: Model Optimization Techniques for Deep Learning & LLM Model
● Quantization (Linear Quantization, Quantization Aware Training (QAT) ,
Post Training Quantization (PTQ) , 1.58-Bit LLMs )
● Knowledge Distillation (Teacher-Student Training)
● Parameter-Efficient Fine-Tuning (PEFT): LoRA(Low-Rank Adaptation),
QLoRA (Quantized LoRA)
● Implementation with Python
Class 23: Reinforcement Learning Intro & LLM Improvement with RAG & RL
● Introduction to Reinforcement Learning (Agent, Environment, Reward)
● LLM Improvement with RAG
● Preference Alignment of LLMs (Reinforcement Learning from Human
Feedback using PPO (Proximal Policy Optimization), Direct Preference
Optimization, Offline RL with Preference Optimization)
● Implementation with Python
Class 24: End-to-End Chatbot Development (Generative AI Project)
● Project Name: End-to-End LLM powered Chatbot with Ollama,
Langchain, Vector Database with ChatUI
Class 25: Advanced GRPO and DeepSeek LLM Underlying Technology
● GRPO: Extending PPO for LLM Training and Optimization
● DeepSeek Architecture and Coder Capabilities
● Efficient Fine-tuning with Unsloth:
○ FlashAttention, Quantization, Memory Optimization
○ Speed Improvements and Benchmarks
● Domain Applications with Implementation Examples
Class 26: AI Agent Workflows with LangChain and CrewAI
● Agent Components: Memory, Planning, ReAct Framework, Tool Use
● LangChain Implementation: Tools, Chains, Memory Systems
● CrewAI for Multi-Agent Systems: Roles, Task Delegation, Communication
● Project: Collaborative Research System Implementation
[Link]
Class 27: Job & Final Project Guidelines
● Resume Building and Portfolio Development (Showcasing Projects and
Skills & Final Project Guidelines)
● ML Industry Interview Guidelines.
Contact Details:
Sohan Khan
Course Coordinator, aiQuest Intelligence & Study Mart
Cell: +8801704265972 (Call/WhatsApp)
Facebook Community: Join Our Community!
Visit Our Pages: Study Mart, aiQuest Intelligence
Watch Free Courses: [Link]
[Link]
