Tirth Bodawala
Setting Up and Testing an HTTP/1.1 Server in Node.js
Build, inspect and test an HTTP/1.1 server in Node.js with only core modules, then fix the shutdown that never exits. Measured on Node v24.21.0.
Read the postFirst principles, priced
Systems & Architecture
First principles with a price on the trade-off: Unix time, CSPRNGs, and Rust in the kernel.
Systems and architecture articles from Atyantik that go to first principles and price the trade-off. We derive Unix time and its Y2038 overflow, separate a statistical PRNG from a kernel CSPRNG by the one question that decides each call, and put a real cost ledger under Rust in the Linux kernel. An automotive set traces one ASIL D function through the V-model and MIL to HIL testing, and a Cloudflare guide decides SSR versus SSG versus edge rendering on Workers.
Filter
Latest posts
Tirth Bodawala
Edge vs origin: count the sequential round trips
Edge vs origin is settled by one countable property: the sequential round trips a request makes to its data. Here is the arithmetic and the counting rule.
Read the post
Tirth Bodawala
Rendering a React app on Cloudflare: SSR, SSG, and edge rendering on Workers
A decision guide for rendering React on Cloudflare Workers: when to prerender to static HTML (SSG), when to run per-request SSR in a Worker, when edge rendering earns its cost, and when to skip SSR entirely. With real Astro adapter and wrangler config.
Read the post
Kalpesh Patel
AI Programming Languages: Pick One Per Stack Layer
You don't pick one of the top AI programming languages. You design a four-layer stack where each layer chooses its own, against your real constraints.
Read the post
Ayush Makwana
How computers generate random numbers: PRNG vs CSPRNG
How computers generate random numbers: statistical PRNGs versus kernel CSPRNGs, the one question deciding each call, and why the wrong pick predicts tokens.
Read the post
Ayush Makwana
Near-data processing: the break-even in bytes
Near-data processing, quantified: computational storage and eBPF share one break-even. It pays below a selectivity crossover you can compute.
Read the post
Ayush Makwana
Unix time epoch explained: epoch, Y2038, leap seconds
Unix time epoch explained by derivation: the one invariant, epoch and timezone conversion, the Y2038 overflow, leap seconds, monotonic vs wall clock.
Read the post
Ajay Patel
Automotive software testing techniques: MIL, SIL, HIL
Automotive software testing techniques as a defect-escape ledger: what MIL, SIL, PIL, HIL, and on-vehicle each catch, are blind to, and cost, traced through one ASIL D requirement.
Read the post
Ajay Patel
Automotive software development process: V-model to SOP
The automotive software development process as one program spine: V-model, ASPICE, ISO 26262, MISRA, and AUTOSAR clipped to one ASIL D steering function.
Read the post
Ajay Patel
What is automotive software: domains, E/E, and the SDV
What is automotive software? A three-axis anatomy: functional domain, E/E architecture location, and software-stack layer, with a live feature locator.
Read the post
Tirth Bodawala
Rust for Linux kernel: the TCO of memory safety
Rust for Linux kernel work, priced both ways: the energy tax of the managed-runtime alternative, the patch cycles a CVE class removes, and an honest cost ledger.
Read the post
Tirth Bodawala
USB Printer Connection Guide: Zebra & Printronix
Troubleshoot Zebra and Printronix USB printer connections on Windows. Learn how Device Interface GUIDs, symbolic links, and USBPRINT device identification solve compatibility issues.
Read the post