YouTube Video Curriculum & Transcripts
Lectures, architectural walkthroughs, and educational masterclasses published on the official YouTube channel (our recorded series). Every recorded session is transcribed in full below with lesson notes, code samples, and mathematical proofs.
Bhāva Saṃskṛta: Computing, Grammar & Architecture
The channel explores the deep structural parallels between ancient Sanskrit generative grammar (Panini's Ashtadhyayi) and modern computer science: context-free grammars, deterministic finite automata, compiler design, low-latency memory management, and capital-efficient venture building.
Video Masterclasses & Annotated Transcripts
How Ancient Sanskrit Grammar Anticipated Computer State Machines
Panini composed the Ashtadhyayi around 500 BCE: a generative grammar containing 3,959 algebraic sutras that function exactly like modern rewrite rules in Backus-Naur Form (BNF). This lecture demonstrates how Sanskrit grammatical derivation maps 1-to-1 onto finite state automata, lexer tokenization, and deterministic pushdown parsers.
real-time In-Memory Architecture: Zero-Disk Verification at Scale
The physics of hardware: RAM accesses take 50 to 100 nanoseconds, while NVMe disk seeks take 10 to 50 microseconds (500 times slower). This session deconstructs the Fraoula in-memory validation engine, circular ring buffers, zero-copy socket parsing, and cryptographic audit proofs that process billions of events without persistent disk logging.
The $12M Cloud Arbitrage: Eliminating Enterprise Server Waste
A comprehensive walkthrough of the FinOps audit framework executed at CAI. Arkajit explains how automated instance shutdown schedules, rightsizing over-provisioned virtual machines, deleting unattached disk volumes, and negotiating enterprise reserved instances recovered 12 million dollars in annual operating margins.
Hyper-Lean Engineering Topology: How 10 Builders Outship 100
Brooks' Law proved that adding people to a late project makes it later. In this lecture, Arkajit calculates communication overhead using the n*(n-1)/2 formula and demonstrates why a 5-to-10 person autonomous engineering pod with end-to-end context ships higher quality software in weeks than a matrixed corporate hierarchy does in quarters.
Request a Topic or Submit an Architecture for Video Review
Subscribers and engineering teams are invited to submit architecture diagrams, cloud bill breakdowns, or compiler design questions for future video teardowns on the channel.