Medium

Codehimanshu

he/him

Codehimanshu

·

1d ago

Two Neurons Out of 295,000: I Changed an AI’s Personality Without Retraining It

Two neurons out of 295,000. No retraining, no prompt. What happened when we edited a mind made of numbers, and why sarcasm turned out to be…

Two Neurons Out of 295,000: I Changed an AI’s Personality Without Retraining It

Codehimanshu

·

Sep 27

The AI Race Has a Power Bill. Who Gets the Future?

From Taiwan’s chip factories to India’s hiring data, the future of AI is being decided far beyond the chatbot window.

The AI Race Has a Power Bill. Who Gets the Future?

Codehimanshu

·

Sep 21

Fast, Calibrated Decisions: What RLCD Models Actually Are, Where They Fit, and Why We Need Both…

Inside TypeSafe Jev, ModernBERT Laya, Robotics Reflexes, and the Multimodal Future of AI Control Flow

Fast, Calibrated Decisions: What RLCD Models Actually Are, Where They Fit, and Why We Need Both…

Codehimanshu

·

Sep 20

I Built a Knowledge Graph of the 4 Vedas: 165,000 Nodes, Half a Million Connections, and the Deep…

How 20,210 sacred mantras across four recensions became a 510,000-edge graph and the engineering story of what it took to build it.

I Built a Knowledge Graph of the 4 Vedas: 165,000 Nodes, Half a Million Connections, and the Deep…

Codehimanshu

·

Aug 31

Building a Post-Apocalyptic Agentic AI Framework on My Old Laptop

How I built an offline system with local models, retrieval, memory, tools, guardrails, and no cloud.

Building a Post-Apocalyptic Agentic AI Framework on My Old Laptop

Codehimanshu

·

Jul 5

Your Codebase Can’t Talk. That’s Why Your Agents Can’t Think.

Loop engineering and harness engineering are the hot new disciplines in agentic coding. Neither one works on a codebase that was never…

Your Codebase Can’t Talk. That’s Why Your Agents Can’t Think.

Codehimanshu

·

May 16

Generative Models vs World Models: The Next Battle for Intelligence

What if today’s AI is only dreaming, while the next generation will actually understand reality?

Generative Models vs World Models: The Next Battle for Intelligence

Codehimanshu

·

Apr 5

The Missing Foundation of Agentic AI: Guardrails, Audit Trails, and Trust

We’re building powerful AI systems but can we actually trust what they do?

The Missing Foundation of Agentic AI: Guardrails, Audit Trails, and Trust

Codehimanshu

·

Mar 1

How I Built a Knowledge Graph of the Ramayana (and Made an AI That Actually Understands It)

From 18,632 ancient verses to 20,000+ nodes, 30,000+ relationships, and an AI that can trace Rama’s family tree in milliseconds here’s the…

How I Built a Knowledge Graph of the Ramayana (and Made an AI That Actually Understands It)

Codehimanshu

·

Jan 26

Building My First Agentic Trading System for the Indian NSE

Understanding real-time agents by throwing them into the most unforgiving environment I know: the stock market.

Building My First Agentic Trading System for the Indian NSE

Codehimanshu

Codehimanshuhe/him

I make ideas into reality. GitHub : https://github.com/HimanshuMohanty-Git24 LinkedIn :https://www.linkedin.com/in/himanshumohanty/ portfolio: himabuilds.in

Following

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech