Data Structures, Algorithms & AI-Era Foundations

Full course walkthrough: Algorithms & Data Structures

How to recognize which pattern a problem needs

Big O notation — how to actually reason about it

Arrays, hash maps & linked lists

Mock interview loop and feedback rubric

Two pointers: converging, parallel, and fast/slow variants

Hash maps and sets: turning O(n^2) into O(n)

Linked lists: pointer rewiring, dummy heads, and the classic operations

Fast & slow pointers: cycle detection and finding the middle

Sliding windows: fixed-size and variable-size

Binary search beyond a plain sorted array

Stacks, queues & the monotonic stack pattern

Heaps, the top-K pattern & priority queues

Merging, inserting & sweeping over intervals

Prefix sums: answering range-sum queries in O(1)

Binary trees & BST operations

Tries: prefix trees for fast string lookups

Graph representations and traversal algorithms

Backtracking: explore, recurse, un-choose

Memoization, tabulation & the classic DP patterns

Greedy algorithms: when the locally-best choice is globally best

Sorting algorithms, and sorting as a preprocessing step

Bit manipulation: the operators and the tricks they enable

Math & geometry patterns that show up in interviews

The pattern cheat sheet

A problem solved end to end

Your interview study plan

Easy set — five warm-ups with full solutions

Medium set — the interview bread and butter

Hard set — when the follow-up gets brutal

What changes in the AI era, and what stays

Vectors, the dot product, and why embeddings work

Gradient descent, and every optimizer built on it

Forward pass, loss, and backpropagation

Self-attention, multi-head attention, and causal masking

BPE tokenization and the embedding table

From cosine similarity to a working RAG pipeline

From logits to the next token

Next-token prediction and the pretraining loop

Adapting a base model: full fine-tuning, feature extraction, instruction tuning

LoRA: train a low-rank delta instead of the whole weight matrix

From instruction-tuned to aligned: reward models, PPO, and DPO

Sinusoidal, learned, RoPE, and ALiBi

Convolution, pooling, receptive fields, and residual connections

Compress to a latent, reconstruct — and make the latent a distribution

The generator vs discriminator game

Add noise, then learn to remove it

CLIP, shared embedding spaces, and how vision reaches an LLM

A*, beam search, alpha-beta, and Monte Carlo Tree Search

The agent loop: reason, act, observe, and the pieces around it

PageRank, and message passing in graph neural networks

Regression, KNN, trees, ensembles, SVM, K-means, PCA

Collaborative filtering, matrix factorization, two-tower retrieval, and learning to rank

Running big models in 8 and 4 bits

Making a model smaller: remove weights, or teach a student

Making LLM generation fast

Data, tensor, and pipeline parallelism, and how gradients sync

Precision/recall/F1, ROC-AUC, perplexity, NDCG, and RAG faithfulness