Building things, breaking things, and occasionally understanding them from Kerala, India.
I'm a developer who thrives on turning random ideas into scalable, working projects. I'm deeply interested in the intersection of AI, backend infrastructure, and modern web development.
- 🤖 Exploring: AI-powered applications and advanced prompt engineering
- 🌐 Building: Efficient web apps, robust backend systems, and developer tools
- 🧠 Learning: System design, scalability, and something new every day
- ⚡ Focus: Practical, high-utility projects that solve actual problems
Frontend
Backend & APIs
Database & Infrastructure
Focused on building efficient, developer-first tools with real-world utility.
A self-hosted, browser-based workspace designed to centralize developer tools and workflows into a single, accessible interface. Built for simplicity, speed, and extensibility.
Highlights:
- 🌐 Fully browser-accessible developer environment
- ⚡ Lightweight architecture with minimal overhead
- 🧩 Designed for extensibility and modular tool integration
Tech Stack: TypeScript Node.js HTML CSS
A minimal yet powerful framework for building interactive terminal applications. Klix focuses on clean abstractions, session-driven architecture, and developer ergonomics.
Highlights:
- 🖥️ Rich terminal UI with structured session handling
- 🔄 Lifecycle hooks and state management built-in
- 🧱 Modular command system for scalable CLI apps
Tech Stack: Python prompt_toolkit Rich
A streamlined Python framework for building Telegram bots with modern async capabilities and efficient networking.
Highlights:
- 🤖 Simple and clean bot development workflow
- ⚡ Async-first design for performance
- 🌐 HTTP/2 support for faster API communication
Tech Stack: Python httpx
Production-ready spam detection with intelligent 6-category classification. Trained on 158.6K multilingual messages with ONNX-powered inference.
Highlights:
- 🛡️ Binary spam detection + 6-category classification (Phishing, Job Scams, Crypto, Adult, Giveaway, Marketing)
- 🌍 8 languages supported (English, Spanish, Chinese, Arabic, Hindi, German, Russian, French)
- ⚡ Ultra-lightweight inference (<15MB RAM, <5ms latency)
- 🎯 93%+ accuracy with ONNX-powered models
- 📊 Trained on 158.6K curated + synthetic messages
- 🚀 Production-deployed in numerous moderation systems
Resources:
Tech Stack: Python Scikit-learn ONNX Runtime TF-IDF Logistic Regression