Tagmind — Tag-Based Notes with LLM-Assisted Organization
A tag-based note app for Android with lightweight capture, LLM-powered search, and encrypted local storage. Currently in closed beta on Google Play — here's how to join and try it.
Recent Posts
Breaking Down the Graph Family: Course Schedule to Clone Graph to Word Ladder
Kahn's algorithm for detecting cycles, a dictionary that doubles as both a visited-check and a lookup table, and a word game that turns out to be BFS in disguise — all reusing the same snapshot-queue engine from grid BFS.
Graph Day: Reusing the Same Engines Again
Course Schedule, Clone Graph, and Word Ladder — three very different problems, but each one reused an engine I already built for something else.
Breaking Down the Tree Family: Level Order to Validate BST to LCA
Grid BFS's snapshot trick reused on trees, four ways to validate a BST, and a lowest-common-ancestor algorithm that turns out to be the same trick as linked list intersection wearing a different hat.
Tree Day: Same BFS Engine, a New Kind of Ancestor Chase
Level Order reused grid BFS's snapshot trick outright. Validate BST needed a range from every ancestor, not just the parent. And LCA with parent pointers turned out to be linked list intersection in disguise.
Meeting Rooms III: Two Heaps, One Delayed Start
Meeting Rooms III needs two heaps instead of one, because now both 'is a room free' and 'which specific room' matter at the same time. A delayed meeting keeps its original duration, not its original end time.
Car Pooling: Weighted Sweep, Bounded Coordinates, and Knowing Who Broke Capacity
Car Pooling is Meeting Rooms II with weighted events. Three ways in: a bucket array when stations are bounded, sorted events when they're not, and a heap when I need to know which trip caused the overflow.
Breaking Down Grid BFS
Four problems, one skeleton — Number of Islands, Max Area, Rotting Oranges, and Flood Fill. Learn the skeleton once, then track only what changes.
Breaking Down the Interval Family: Merge Intervals to Meeting Rooms III
One family, one first move — sort by start. Then it branches: merge into blobs, count with a sweep or a bucket array, keep room identities with a heap, add passenger weights, or juggle two heaps when a delayed meeting needs a specific room. The question that splits every follow-up: do you need WHO, or just HOW MANY?
Breaking Down the Stack Family: Valid Parentheses to LRU Cache
Counters vs stacks, why Min Remove stores indexes, Min Stack three ways, and the design story behind LRU Cache — a hashmap that answers 'where' married to a linked list that answers 'who's oldest'.
Interval Day: One Sweep, Three Ways to Line Up Time
Merge Intervals, Insert Interval, and Meeting Rooms II solved three ways — heap, two pointers, and a bucket array. Plus the question that splits them all: do I need to know WHICH room, or just HOW MANY?