Design Gurus — searchable YouTube transcripts & AI Q&A

How Modern Login Systems Work: SSO, OAuth, OIDC & SAML

Modern login systems separate authentication, authorization, and single sign-on, with SSO as the user experience, OAuth 2.0 for delegated authorization, OIDC for authentication built on OAuth, and SAML for enterprise federation. The speaker emphasizes validating tokens and assertions correctly, using authorization code with PKCE, and choosing OIDC for modern apps, OAuth for API access, and SAML for enterprise integrations.

PostgreSQL vs MySQL: The Differences Developers Miss

PostgreSQL and MySQL are both powerful relational databases, but their architectural differences matter at scale. PostgreSQL offers more flexibility with specialized indexes, JSONB, and extensions like PostGIS and pgvector, while MySQL's InnoDB provides efficient primary key lookups and a mature ecosystem. The right choice depends on your workload, team expertise, and operational constraints, not on a feature checklist.

Your SQL Queries Are Slow? Learn These 7 Database Indexes

Database indexes are specialized structures that solve different query problems, and choosing the wrong type can waste storage or slow down writes. The speaker explains that B-tree indexes are best for general-purpose lookups and range queries, while hash indexes suit exact-match searches, and composite indexes should be designed around actual query patterns. He also notes that more indexes don't always mean better performance, and recommends using EXPLAIN to inspect execution plans before adding indexes.

SQL Query Execution Explained: Parsing, Optimization & Indexes

SQL queries are processed in four stages: parsing, optimization, execution, and returning results, with the optimizer choosing an execution plan based on estimates that may not always be fastest. Using EXPLAIN (or EXPLAIN ANALYZE in PostgreSQL) reveals the actual plan, helping identify expensive operations like sequential scans versus index scans. Understanding this process is key to writing faster SQL queries.

Why Your Vibe-Coded App Will Crash at 10,000 Users

Vibe-coded apps often crash under real traffic because AI-generated code lacks system design considerations like database indexes, caching, and stateless servers. The speaker, a former Meta and Microsoft engineer, argues that developers must add indexes on queried columns, avoid N+1 queries, use caching, offload slow tasks to queues, and protect paid API keys to survive viral growth. He emphasizes that AI writes code, but humans must design the system to handle scale.