Archives

My local voice-to-text app sometimes pasted hundreds of repeated words. The fix was VAD-aware audio clipping plus a conservative text filter, not another language model.

Codex CLI warns that skill descriptions were shortened to fit the skills context budget. Here is what it costs you and how to get back under the limit.

The integrated terminal renders ghost characters, blank rows, or a smeared cursor. One setting fixes it.

People ask how one person keeps a channel moving, so I made a video walking through my whole setup. Here is the gist. The boring parts stopped being manual. They became custom skills inside a seven-stage Claude Code pipeline that runs from my terminal: Strategy, Ideate, Produce,

OpenAI shipped `/goal` in the Codex CLI and the Codex app, so I made a video showing what it does and how to turn it on, since that is the question I kept seeing in every thread. The pitch: set a task, walk away, come back hours later to find it done, no babysitting the agent eve

I hit my Codex weekly limit with three days still left in the week, which used to mean waiting it out. OpenAI added a way to reset the limit yourself, so I made a short video showing it live. Here are the exact steps if you would rather read them than watch. You no longer have to

If you want Claude Code but do not want to burn API credits, Google made that easier. I made a video installing Gemini CLI from scratch and putting it through a real task, and here is the gist plus the steps so you can follow along without watching if you prefer. It is an open-so

I set myself a small challenge: explain generative AI in about 90 seconds, no jargon, one analogy. The final cut runs about 81 seconds. Here's the idea behind it.

The Kimi K3 setup for Hermes Agent via OpenRouter is two files. Here is the walkthrough, the copy-paste configs, and the token-cost catch to test before you commit.

I made a quick video on how I stopped drawing architecture diagrams by hand, and this post is the written version so you can copy the workflow straight off the page. I used to lose half an hour dragging boxes around in Lucidchart or Figma just to draw one diagram. Now I describe

I made a quick video on a Codex setting most people never open, because flipping it on changed how I run bigger tasks. Multi-agent mode lets Codex run several agents at once, each with its own context window, reporting back to the main agent. In the video I turn it on and push it

Using Claude Code and Claude Web to analyze my iCloud folders, find bloat, and cut my monthly storage bill from $9.99 to $2.99.

How I optimized my Hexo blog from a Lighthouse score of 79 to 94 by automating image compression (WebP), enabling lazy loading for images and scripts, and configuring proper cache headers.

A deep dive into the modern serverless tech stack (Hexo, AWS S3, CloudFront) and the AI-first workflow used to build and maintain this blog.

How I consolidated ai.saurav.io and cloud.saurav.io into a single domain using AWS CloudFront Functions for managing 301 redirects at the edge.

Comparing the peak hype cycles of Cloud Computing vs. AI: why AI's direct user impact and media speculation drive even higher public engagement.

Why User Experience (UX) is the key differentiator for AI applications, exploring how great design separates successful AI tools from the rest.

Why Human-in-the-Loop (HITL) workflows are evolving from simple safety checks to strategic advantages in the era of sophisticated AI agents.

A short reflection on the rise of LLM-based AI Agents and the prediction of a future populated by trillions of autonomous digital workers.

Streamlining Cloud App Development by combining ChatGPT's code generation with the Serverless Framework to instantly deploy a Python-based DynamoDB API.

A roundup of essential AI tools for creators and developers, including ChatGPT, Midjourney, Eleven Labs, and RunWayML.

Exploring early AI image generation tools like DALL-E 2, Stable Diffusion (DiffusionBee), and Midjourney, and their rapid evolution.

Troubleshooting notes for common AWS CDK Workshop issues, including CLI version mismatches, TypeScript configuration errors, and CodeCommit branch matching.

How to use AWS CloudFormation tag inheritance to automatically propagate tags to all supported resources in a stack for better cost allocation and management.

Drawing analogies between AWS CloudFormation and Soccer: comparing stacks to games, resource limits to player rosters, and updates to substitutions.

The first post on my cloud engineering blog. Why I decided to start writing about AWS, cloud architecture, and the technologies shaping modern infrastructure.