$ npm run build
 
compiled successfully in 312ms
main build passing v1.0.0 · Bhilai, IN

~/home.tsx

~/projects.json

Changelog

A running log of things I've shipped — framed the way I actually think about them: as releases.

agent

Blog-Writing-Agent

Python

This agent doesn't just "write" — it thinks. Utilizing a multi-step workflow, the agent performs real-time web research, scrapes relevant data, constructs a semantic outline, and drafts high-quality blog posts autonomously.

  • Added real-time web research and semantic scraping pipeline
  • Added multi-step outline construction before drafting
  • Added autonomous high-quality blog generation workflow
agent

React-Github-Agent

Python

A production-ready autonomous AI agent powered by LangGraph, FastMCP, and dual-memory to seamlessly manage, develop, and interact with your GitHub repositories.

  • Added dual-memory architecture for persistent context across sessions
  • Added FastMCP integration for GitHub repository management
  • Added LangGraph-powered autonomous task execution loop
infra

ai-failover

HTML

Demand-aware AI API router that automatically fails over to a backup provider when traffic exceeds a threshold — inspired by Gemini's high-demand behaviour.

  • Added demand-aware traffic threshold detection
  • Added automatic failover to backup AI provider
  • Added real-time provider health monitoring
systems

AI-System-Design

Docs

A collection of system design documents for production AI systems — focused on architecture decisions, tradeoffs, and design patterns. No code, just thinking.

  • Added architecture decision records for production AI systems
  • Added tradeoff analyses and design pattern documentation
cli

groq-cli-agent

TypeScript

A TypeScript CLI AI agent powered by Groq that interacts with the filesystem using natural language commands.

  • Added natural language filesystem interaction via Groq
  • Added TypeScript CLI interface with rich terminal output
backend

ts-auth-jwt

TypeScript

A TypeScript REST API with JWT authentication and role-based authorization using Express, bcrypt, and jsonwebtoken.

  • Added JWT authentication with role-based access control
  • Added bcrypt password hashing and secure token refresh flow
  • Added 401 vs 403 middleware distinction for auth errors

~/about.md

About

I'm a second-year B.Tech student in Data Science & AI at IIT Bhilai. I got into AI by wanting to understand how language models actually work under the hood — not just call APIs, but really understand the internals. That curiosity has stuck.

These days I split my time between research (NLP, mechanistic interpretability) and building systems — agents, RAG pipelines, inference platforms, Android apps. I like building things from scratch because that's the fastest way to actually learn them.

Outside of code I'm usually reading about how the brain works, or finding ways to apply those ideas to make AI systems smarter.

// currently

  • Undergraduate researcher @ IIT Bhilai
  • Building an AST-based Python chunking library for coding agents
  • Studying LLM internals: MoE, FlashAttention, MLA
  • Learning Reasoning models, World models & AI Agents
  • Future: AI × Cybersecurity & Quantum computing
  • Open to research collabs & internships

~/requirements.txt

Dependencies

What I actually use — from model training to deployment. A few things I've happily moved on from.

{
  "languages": {
    "python": "^3.11.0",
    "typescript": "^5.4.0"
  },
  "ai_ml": {
    "pytorch": "^2.3.0",
    "langchain": "^0.2.0",
    "langgraph": "^0.1.0",
    "huggingface_hub": "^0.23.0",
    "groq": "^0.9.0",
    "ollama": "^0.2.0"
  },
  "backend_infra": {
    "fastapi": "^0.111.0",
    "node": "^20.0.0",
    "docker": "^26.0.0",
    "postgresql": "^16.0.0",
    "redis": "^7.0.0",
    "qdrant": "^1.9.0"
  },
  "devDependencies": {
    "curiosity": "^∞.0.0",
    "first-principles-thinking": "^1.0.0"
  },
  // retired, no hard feelings
  "deprecated": [
    "black-box-thinking",
    "tutorial-hell"
  ]
}

~/contact.sh

Get in touch

$
Status: open to opportunities · avg. reply time < 24h