Q3 2025 Action Plan

Summer 2025 Portfolio Update

Phase 1: Foundation

Update Action Plans UI

Create task lists Q3 2025

Migrate stories to client-rendered format (e.g. dynamic import)

Write and add new stories

Categorize projects: frontend, backend, full sites, tools, web & mobile apps

Redesign project section with filtering and search

Phase 2: Content & Credentials

Add all published research with title, abstract, link, and thumbnail

Add diploma with PDF/image and badge if possible

Design CV page and downloadable PDF

Highlight CV sections: education, experience, projects, research, skills

Phase 3: Showcase & Expansion

Create startup landing page: problem, solution, features, stack, roadmap

Design the page for investors and collaborators (clean and persuasive)

Phase 4: Iteration & Polishing

Review performance (Core Web Vitals, lazy loading, images)

Optimize SEO (meta tags, structured data for projects/stories)

Add optional blog or insight section (tech tips, reflections)

Polish UI: micro animations, dark mode, responsive UX

Deep Learning - D2L.ai Book

Preface & Setup

Read Preface

Installation

Understand Notation

1. Introduction

Introduction

2. Preliminaries

Data Manipulation

Data Preprocessing

Linear Algebra

Calculus

Automatic Differentiation

Probability and Statistics

3. Linear Neural Networks for Regression

Linear Regression

Object-Oriented Design for Implementation

Synthetic Regression Data

Implementation from Scratch

Concise Implementation

Generalization

Weight Decay

4. Linear Neural Networks for Classification

Softmax Regression

Image Classification Dataset

Base Classification Model

Implementation from Scratch

Concise Implementation

Generalization in Classification

Environment and Distribution Shift

5. Multilayer Perceptrons

MLP Basics

Implementation

Forward & Backward Propagation

Numerical Stability & Initialization

Generalization in Deep Learning

Dropout

Predicting House Prices on Kaggle

6. Builders' Guide

Layers and Modules

Parameter Management

Parameter Initialization

Lazy Initialization

Custom Layers

File I/O

GPUs

7. Convolutional Neural Networks

From Fully Connected to Convolutions

Convolutions for Images

Padding and Stride

Multiple Input & Output Channels

Pooling

LeNet

8. Modern CNNs

AlexNet

VGG

Network in Network (NiN)

GoogLeNet

Batch Normalization

ResNet & ResNeXt

DenseNet

Designing CNN Architectures

9. Recurrent Neural Networks

Working with Sequences

Converting Raw Text

Language Models

RNN Basics

RNN Implementation from Scratch

Concise RNN Implementation

Backpropagation Through Time

10. Modern RNNs

LSTM

GRU

Deep RNNs

Bidirectional RNNs

Machine Translation Dataset

Encoder-Decoder Architecture

Seq2Seq Learning

Beam Search

11. Attention Mechanisms & Transformers

Queries, Keys, and Values

Attention Pooling

Attention Scoring Functions

Bahdanau Attention

Multi-Head Attention

Self-Attention & Positional Encoding

Transformer Architecture

Vision Transformers

Large-Scale Pretraining with Transformers

Previous Action PlanNext Action Plan