अभ्यास • Week 22

verified_user

Weekly Production Deliverable

Target Output

Deliverable: Binary Search Trees fully done. Creational Design Patterns done. You've built a GPT and a BPE tokenizer from scratch — you understand exactly what happens inside every LLM API call you'll ever make.

Weeks: W21 W22 Active W23 W24 W25

Monday

0 / 2 done

DSA Introduction to BST; Search in BST

1.5h

AI/ML Karpathy: "Let's build GPT" Part 1 (Self-Attention)

Andrej Karpathy — "Let's build GPT: from scratch, in code, spelled out": Self-Attention mechanism from first principles, Query, Key, Value matrix math, Scaled dot-product attention, causal masking for autoregression

2.5h

Tuesday

0 / 2 done

DSA Floor and Ceil in a BST; Insert a given node in BST

1.5h

CORE Creational Design Patterns 1

Singleton Pattern (Thread-safe Eager, Lazy, Double-Checked Locking, Bill Pugh, Enum) & Factory Method Pattern + Java code implementations + Quiz

2.5h

Wednesday

0 / 2 done

DSA Delete a node in BST; Kth Smallest and Largest element in BST

1.5h

AI/ML Karpathy: "Let's build GPT" Part 2 (Full Transformer Block)

Multi-Head Attention, Feedforward networks, Residual connections, Layer Normalization, Dropout; assemble the full decoder-only transformer block

2.5h

Thursday

0 / 2 done

DSA Check if a tree is a BST or not; LCA in BST LC #A

1.5h

CORE Creational Design Patterns 2

Abstract Factory Pattern, Builder Pattern (Fluent API, telescoping constructor alternative), Prototype Pattern (Shallow vs Deep cloning) + Java code implementations + Quiz

2.5h

Friday

0 / 2 done

DSA Construct a BST from a preorder traversal; Inorder successor and predecessor in BST

1.5h

AI/ML Karpathy: Build the GPT Tokenizer & nanoGPT Release

Karpathy: "Let's build the GPT Tokenizer" (Byte Pair Encoding - BPE algorithm from scratch, training vocabulary, encoding/decoding text) + push your completed nanoGPT + BPE Tokenizer repo to GitHub with README

2.5h

⚡ Weekend High-Load Execution • 8.0h Daily Deep Focus Lab

Saturday

0 / 3 done

DSA BST iterator; Two sum in BST

1.5h

AI/ML Self-Attention & Scaled Dot-Product Attention from Scratch (4.0h)

Code multi-head self-attention mechanism from scratch in PyTorch with Query, Key, Value projections and causal masking.

4.0h

APT Quant M22 & Verbal/Logical

Quant M22 Probability (Basic & Advance) + Verbal/Logical Mixed Practice

2.5h

Sunday

0 / 3 done

DSA Correct BST with two nodes swapped; Largest BST in Binary Tree + Contest + full BST revision

1.5h

AI/ML "Attention Is All You Need" Landmark Paper Read (4.0h)

Vaswani et al. (2017) full paper analysis; annotate transformer architecture diagram, positional encodings, and feed-forward sublayers.

4.0h

APT Creational Patterns Mastery

Implement 2 Creational Patterns from scratch in Java + explain Self-Attention and BPE tokenization out loud using your own code

2.5h

edit_note

Week 22 Technical Notes & Journal

#LeetCode #Java #DSA #Week22