rahul--pandit

geek notes and random thoughts

AI-First SEO, AEO, GEO, and RAO

AI-First SEO, AEO, GEO, and RAO

Optimization for the AI-Driven Web

rahul--pandit.bsky.social

MarTech - 101 Explained

MarTech - 101 Explained

Marketing Technology Essentials

rahul--pandit.bsky.social

Beyond the CWS Dashboard: A Guide to Advanced Chrome Extension Analytics and Localized SEO

Beyond the CWS Dashboard: A Guide to Advanced Chrome Extension Analytics and Localized SEO

For Chrome extension developers, understanding performance and user retention is critical for growth. While the official

rahul--pandit.bsky.social

How To Use Kalman Filters To Build Adaptive Pairs Trading Strategies (Quant Framework)

How To Use Kalman Filters To Build Adaptive Pairs Trading Strategies (Quant Framework)

Static hedge ratios kill pairs trades. You compute a 60-day rolling regression, get a hedge ratio of 0.72, and trade the spread. Two weeks later the fundamental relationship has shifted - the true hedge ratio is now 0.85 - but you're still trading at 0.72. You're bleeding.

rahul--pandit.bsky.social

There Are Only 4 Ways to Make Money. Everything Else Is a Story.

There Are Only 4 Ways to Make Money. Everything Else Is a Story.

Let me ruin a few things you believe about money.

rahul--pandit.bsky.social

Smart doorbell without any subscription

Smart doorbell without any subscription

Off-the-shelf smart doorbells, such as those from Ring, require you to pay monthly subscription fees. On top of that, your footage is beamed to Amazon's servers, which can be problematic, given the company's track record of sharing Ring footage with law enforcement.

rahul--pandit.bsky.social

How does Claude's watermarking actually work?

How does Claude's watermarking actually work?

Transcript from CTO of GPTZero https://www.linkedin.com/pulse/how-does-claudes-watermarking-actually-work-alex-cui-v1lee Anthropic dropped a bombshell announcement yesterday and it was through their help center.

rahul--pandit.bsky.social

How To Build a Quant Portfolio From Scratch (2026 Edition)

How To Build a Quant Portfolio From Scratch (2026 Edition)

Not a stock-picking guide. The actual mathematics behind capital allocation from covariance estimation to when you cut a strategy entirely.

rahul--pandit.bsky.social

One Equation Controls Every Dollar You Will Ever Make in Markets.

One Equation Controls Every Dollar You Will Ever Make in Markets.

Knowing what to ignore is a bigger edge than finding more data

rahul--pandit.bsky.social

You Were Taught One Kind of Average. The One That Actually Runs Your Money Is a Different Number.

You Were Taught One Kind of Average. The One That Actually Runs Your Money Is a Different Number.

Here is a bet. You put up 100 dollars. A fair coin decides. Heads, your money grows by 50 percent. Tails, it shrinks by 40 percent. Expected value per round: plus 5 percent. Take the bet, and take it again, and take it forever. What happens?

rahul--pandit.bsky.social

Depression is not a mindset, a bad day, or a lack of strength

Depression is not a mindset, a bad day, or a lack of strength

Robert Sapolsky - neurobiology behind human suffering

rahul--pandit.bsky.social

LLM - Symbolic + Embedded Space

LLM - Symbolic + Embedded Space

Views on AI infrastructure, long-horizon limitations, and enterprise adoption strategies.

rahul--pandit.bsky.social

Loop Engineering

Loop Engineering

Let's take a look at what is Loop Engineering, understand its benefits and address the key structural weaknesses, cost realities, and operational risks of autonomous AI coding loops.

rahul--pandit.bsky.social

How you can use Atlas Browse as an Autonomous Business Operator

How you can use Atlas Browse as an Autonomous Business Operator

Instead of using ChatGPT like a glorified chatbot - typing prompts, copying responses, and pasting text across dozens of open tabs: Move from prompting to delegating.

rahul--pandit.bsky.social

Free, Automated RSS Reader with Liveboat and GitHub Pages

Free, Automated RSS Reader with Liveboat and GitHub Pages

Here is how you can use Liveboat - an open-source tool to completely transform how you consume news.

rahul--pandit.bsky.social

Microsoft SkillOpt: From Prompt Engineering to Skill Training

Microsoft SkillOpt: From Prompt Engineering to Skill Training

A Deep Dive Into SkillOpt’s Framework, Experimental Results, and Open-Source Implementation

rahul--pandit.bsky.social

KV Caching in LLMs, Clearly Explained

KV Caching in LLMs, Clearly Explained

You've seen it every time you use ChatGPT or Claude. The first token takes noticeably longer to appear. Then the rest stream out almost instantly.

rahul--pandit.bsky.social

Google - Introducing the Open Knowledge Format

Google - Introducing the Open Knowledge Format

Google recently introduced the Open Knowledge Format (OKF), an open specification that formalizes the LLM-wiki pattern into a portable, interoperable format. This is a vendor-neutral, agent- and human-friendly standard for representing the metadata, context, and curated knowledge that modern AI systems need.

rahul--pandit.bsky.social

Percentiles Explained

Percentiles Explained

What Is P10, P50, and P90?

rahul--pandit.bsky.social
Percentiles Explained

A Beginner's Guide to Robotics Hardware

A Beginner's Guide to Robotics Hardware

https://interlatent.com/blog/interlatent-robotics-hardware-guide

rahul--pandit.bsky.social

Beyond Prompting: How to Actually Engineer AI Workflows into Business Systems

Beyond Prompting: How to Actually Engineer AI Workflows into Business Systems

The article argues that the cutting edge of AI adoption has shifted from simple prompting to building "loops" - autonomous or semi-autonomous execution cycles where an AI model acts as a subroutine inside a programmatic wrapper. It outlines a 5-part framework for applying these loops to core revenue-generating business operations (sales, marketing, recruiting) and warns against unmonitored loop execution.

rahul--pandit.bsky.social

How to test a mobile Edge extension on Android

How to test a mobile Edge extension on Android

To test a mobile Edge extension on Android, you will need to use the Microsoft Edge Canary app from the Google Play Store. You can load your custom extension either by enabling experimental flags for direct ID installation or by connecting remotely via USB to a desktop machine for advanced developer inspection.

rahul--pandit.bsky.social

Behind the Prompt: How Production LLM Systems Scale in the Real World

Behind the Prompt: How Production LLM Systems Scale in the Real World

Want to serve flagship large language models to millions of concurrent users?

rahul--pandit.bsky.social

Text Compare Tool - 3 way or 2 way supported

Text Compare Tool - 3 way or 2 way supported

https://www.guideaxon.com/diff-tool.html

rahul--pandit.bsky.social

(S)AGE - (Sovereign) Agent Governed Experience

(S)AGE - (Sovereign) Agent Governed Experience

https://github.com/l33tdawg/sage

rahul--pandit.bsky.social

PageIndex: Document Index for Vectorless, Reasoning-based RAG

PageIndex: Document Index for Vectorless, Reasoning-based RAG

https://github.com/VectifyAI/PageIndex

rahul--pandit.bsky.social

TurboVec - Vector Index

TurboVec - Vector Index

RyanCodrai/turbovec

rahul--pandit.bsky.social

Leverage classifiers for your Agent

Leverage classifiers for your Agent

One of my favorite superpowers of agents is building classifiers. It’s insanely high leverage. Before AI, you needed a year-round team: - 3 ML engineers to build the models - 3 ML infra engineers to scale them up - 2 software engineers to integrate the parts - 1 data scientist to analyze it - 1 PM to manage the product - 0.5 EM to hold it together Now, in minutes, you can have an agent generate a markdown file that classifies inputs, then let agents run continuously against it.

rahul--pandit.bsky.social

Markdown memory system for you and your AI agent

Markdown memory system for you and your AI agent

iwe.md

rahul--pandit.bsky.social

Knowledge Graph Extractor

Knowledge Graph Extractor

Author - hanxiao https://github.com/hanxiao/knowledge-graph-extractor/ It takes a lot of effort to find a question that's non-trivial to a scaffolded LLM yet still answerable. To find those question-answer pairs, I built a knowledge graph extractor where you can throw a corpus at it, and it extracts the entity relations using qwen3.6-35b-a3b-MTP on an L4 at 70 tps (which is really good for such a low-budget GPU). Then I mark out the longest path in the graph and use it to generate challenging question-answer pairs. The idea is to find those genuinely multi-hop fact chains that are verifiable from the corpus, to stress-test the agentic search system.

rahul--pandit.bsky.social

Breaking Free from EAS Cloud: Building and Distributing Expo Apps on Your Own Terms

Breaking Free from EAS Cloud: Building and Distributing Expo Apps on Your Own Terms

Are you locked into Expo Application Services (EAS) Cloud, and will build and deployment costs scale out of control?

rahul--pandit.bsky.social

Open Source Bookmark Manager

Open Source Bookmark Manager

https://www.xbrowsersync.org/

rahul--pandit.bsky.social

What is a GGUF model?

What is a GGUF model?

How does it work?

rahul--pandit.bsky.social

Deploying LLMs in production?

Deploying LLMs in production?

Here's a hands-on guide:

rahul--pandit.bsky.social

Hire the right talent

Hire the right talent

Marc Andreessen on how to get people to join your 3-person startup Marc says founders have two tools at their disposal to win new hires:

rahul--pandit.bsky.social

Are you using Claude or Codex correctly

Are you using Claude or Codex correctly

Do you just install Claude and use simple 1-2 liner prompts? Learn how to write a Skill and use sub-agents and go from using 5% of the capabilities to 100%.

rahul--pandit.bsky.social