Build videos and shorts from the same workflow behind the posts, tools, and prompts on this site.
Subscribe on YouTube →One database. One million users overnight. CPU pegged at 100%, queries timing out, 502s everywhere. This is the exact 10-step playbook for scaling a relational database without losing data, compromising security, or going bankrupt: vertical scaling, query optimization and indexing, caching, read replicas, message queues, partitioning and sharding, polyglot persistence, and security — then the evolution that ties it together. Chapters: 0:00 The viral bottleneck — 1K to 1M users, one database 0:44 Vertical scaling — same code, bigger box (and its ceiling) 1:17 Query optimization — sequential scans vs B-tree indexes 1:58 Caching — Redis, cache-aside, and invalidation 2:38 Read replicas — multiply reads, meet replication lag 3:22 Message queues — decouple writes with Kafka 4:00 Partitioning and sharding — split logs, shard users 4:45 Polyglot persistence — the right store per workload 5:22 Security and compliance — TLS, encryption, GDPR/FERPA 6:04 The evolution — scale when the data demands it #databases #systemdesign #scalability
Watch →What Actually Travels When You Tap to Pay? Your real card number never leaves your pocket. This is tokenization: a disposable code stands in for your card for exactly one purchase, then dies. Chapters: 0:00 The tap — card or phone, what travels? 0:08 Wallet checkout — Paid, number still locked in 0:18 The secure element vault — apps denied 0:28 Token mint — one purchase, one code 0:38 NFC flight — 13.56 MHz, 4 cm of air 0:48 Bank lookup — only the bank holds the key 0:58 Hacker payoff — dead string of garbage 1:08 Zero transmitted — recap 1:14 Follow ArbindBuilds — for engineering explained #taptoPay #nfc #tokenization
Watch →Why does a massive 4K Netflix stream start faster than a tiny text website? 🤯 It’s not faster internet. It’s **latency**. Netflix moves content closer to you using CDNs, Anycast BGP routing, and its own Open Connect servers inside ISP networks. Instead of making your data travel across the internet to one distant server, Netflix can serve it from a nearby edge location. That’s why a heavier video can feel faster than a simple website. In this 60-second engineering breakdown, we explain: → Bandwidth vs latency → Why distance matters → How CDN edge servers work → Anycast BGP routing → Netflix Open Connect → How cache pre-warming makes playback almost instant **Same internet. A very different journey.** Follow **ArbindBuilds** for engineering explained without the fluff. #Netflix #CDN #Latency #Bandwidth #HowTheInternetWorks #Networking #ComputerScience #TechExplained #WebDevelopment #Internet #BGP #Anycast #OpenConnect #SoftwareEngineering #EngineeringExplained #TechTok #YouTubeShorts #Shorts
Watch →When you open an Incognito window, you think you're invisible. But private browsing doesn't hide you from the web it only hides you from your own computer. ### In this 60-second engineering breakdown, we reveal what actually happens inside your browser when you go private: • Why your local RAM session wipes cookies, but your network remains fully exposed. • How unencrypted DNS queries leak every domain you visit to your ISP over UDP Port 53. • How passive device fingerprinting queries your Canvas 2D render hash, WebGL GPU silicon, and AudioContext drift to identify your physical machine with 99.2% accuracy in under 12ms. • What happens when you log into a single account and collapse the sandbox. ### ⏱️ TIMESTAMPS / CHAPTERS: 00:00 The Invisibility Illusion (Local vs Network) 00:10 RAM Temporary Session vs NVMe SSD Storage 00:24 DNS Leakage & Cleartext ISP Logging (Port 53) 00:38 Passive Device Fingerprinting (Canvas, WebGL, Audio) 00:56 The 99.2% Mathematical Entropy Hash (Cross-Session Link) 01:12 Sandbox Collapse on User Login 01:26 The Core Axiom: Local Amnesia, Not a VPN 01:35 ArbindBuilds ### 💬 QUESTION: Did you know websites can fingerprint your hardware even in Incognito mode? What browser or privacy setup do you run? Drop your thoughts in the comments! 🔔 Subscribe to ArbindBuilds for bite-sized engineering breakdowns into how modern tech works under the hood. #Incognito #Cybersecurity #Privacy #Browser #WebDev #SoftwareEngineering #ComputerScience #ArbindBuilds
Watch →What actually happens when a packet vanishes on the internet? 📡⚡ The internet is a chaotic, lossy network of physical wires and fiber optics. Cables get cut, buffers overflow, and millions of packets disappear every single second. Yet your video never crashes and your download never corrupts. Here is the invisible protocol running inside your operating system that makes it possible: TCP. > 🔹 Sequence numbering > 🔹 3x Duplicate ACKs → Fast Retransmit > 🔹 Out-of-order reassembly > 🔹 Congestion backoff ### If you write software, this is what’s keeping your system alive under the hood. 🛠️ Drop a comment: TCP or UDP for real-time multiplayer games? 👇 #computerscience #programming #softwareengineering #networking #tech #arbindbuilds #webdev #coding
Watch →When you click "I'm not a robot", you aren't proving you can click a button. Any automated script can execute button.click() in 0.8 milliseconds. The real test happens in the 100 milliseconds BEFORE the click: cursor trajectory kinematics, physiological hand tremor, browser hardware fingerprinting, and machine learning risk scoring. ### 🧠 HOW IT WORKS UNDER THE HOOD: > 1. Cursor Kinematics: Bots move in Euclidean straight lines with zero acceleration change. Humans obey Fitts's Law, decelerating organically with a 10.4Hz neuromuscular tremor. > 2. Hardware Probes: JavaScript evaluates Canvas 2D sub-pixel anti-aliasing hashes, WebGL GPU shader silicon signatures, AudioContext clock drift, and navigator.webdriver flags. > 3. Network Reputation: Separates cloud datacenter scrapers (AWS ASN 16509) from trusted residential fiber connections using JA4 TLS handshake fingerprints. > 4. Risk Scoring Engine: Ensemble ML models assign an entropy trust score (0.0 to 1.0). High entropy yields an instant silent pass. > 5. The 3x3 Puzzle: Borderline scores trigger image grids (traffic lights, crosswalks) that double as crowdsourced labeled training data for autonomous vehicles. ### 💬 QUESTION: Have you ever failed a CAPTCHA even though you're definitely human? Drop your funniest or most frustrating CAPTCHA story in the comments! 🔔 Subscribe to ArbindBuilds for engineering deep dives into how modern software and systems work under the hood. #CAPTCHA #Cybersecurity #SoftwareEngineering #ComputerScience #MachineLearning #WebDevelopment #ArbindBuilds how captcha works, how does recaptcha know you are human, how captcha knows you are not a robot, recaptcha explained, fitts law cursor tremor, browser fingerprinting, cloudflare turnstile, bot detection algorithms, computer science explained, arbindbuilds, software engineering, web security
Watch →You scan a QR code in seconds. But what actually happens inside your phone before that link appears? It isn't simply “taking a picture.” Your phone captures the image, detects the QR code's finder patterns, figures out its orientation and grid, removes the masking pattern, extracts the encoded bits, and uses Reed-Solomon error correction to recover data that may be partially damaged. And here's the surprising part: The QR code itself doesn't need an internet connection to be decoded. The decoding happens locally on your device. Only after the payload is decoded does your phone need the internet to open an online URL. In this video, we break down the process step by step: ⏱️ TIMESTAMPS 00:00 What really happens when you scan a QR code? 00:07 1. Your camera captures the QR code 00:16 2. Finding the three “eyes” 00:26 3. How the phone finds the QR grid 00:35 4. The hidden XOR mask 00:44 5. Reed-Solomon error correction 00:55 6. Turning patterns into bits 01:04 7. Decoding the actual payload 01:12 Why the QR code itself doesn't need the internet 🧠 UNDER THE HOOD 1. Image capture Your camera captures the QR code as an image. The decoder then looks for the high-contrast structures that make the code recognizable. 2. Finder patterns Those three large squares in the corners aren't decoration. They help the decoder determine the QR code's position, orientation and scale. 3. Timing patterns Alternating black and white modules help the decoder determine the spacing of the QR code's grid. 4. Masking QR codes use mathematical masking patterns to avoid problematic visual patterns. The decoder reverses the selected mask to recover the original data pattern. 5. Error correction QR codes use Reed-Solomon error correction. Depending on the selected error-correction level, the code can recover from a significant amount of damage or obstruction. 6. Bitstream decoding The modules are converted into bits and interpreted according to the QR code's encoding format. 7. Payload Finally, your phone gets the actual payload, such as a URL, text, contact information, Wi-Fi credentials or another piece of data. Only then does the next step happen: if that payload is a URL, your phone can connect to the internet and open it. So the next time you scan a QR code, remember: You're not just taking a picture. Your phone is decoding a small mathematical data structure in real time. 💬 QUESTION: Have you ever scanned a QR code that was scratched, folded or partially covered and somehow it still worked? What piece of everyday technology should we break down next? 🔔 Subscribe to ArbindBuilds for short, visual explanations of the technology and software systems hiding underneath everyday things. #QRCode #HowQRCodeWorks #ComputerScience #TechExplained #Programming #Algorithms #ArbindBuilds
Watch →