"Lazy on platforms. Relentless in the codebase. Less talk, more functions."

I am a self-driven, self-taught engineer who builds realistic, functioning systems. I don't care about buzzwords, and I don't box myself into a single domain. Real-world problems require cross-domain execution—whether that means writing low-level Rust for on-device AI, designing scalable SQL schemas, deploying offline-first mobile apps, or crafting functional UI/UX. I learn what needs to be learned, and I build what needs to be built.

Core Arsenal: Rust C++ Python TypeScript React/Next.js Bun PyTorch llama.cpp IoT/Hardware

These repositories represent actual engineering problems solved across multiple domains—from edge ML and hardware simulations to full-stack EdTech and local privacy-first AI.

• Domain: Edge AI & Systems Programming
• Tech: Rust, llama.cpp, Qwen-3, Termux, Android
• The Build: A production-grade AI assistant running entirely on Android hardware with zero cloud dependency. Implements a ReAct loop, 4-tier memory, and active inference engine. Built to solve the massive privacy vulnerabilities and internet-dependency of modern AI assistants.

• Domain: Full-Stack & System Architecture
• Tech: Next.js, Node.js, MySQL, AI Engine
• The Build: A comprehensive, scalable monorepo for a smart examination platform. Engineered the secure backend, AI integration, and optimized Next.js landing pages to handle end-to-end test management, solving the rigidity of legacy exam platforms.

• Domain: Mobile Automation & ML
• Tech: Sarvam-1 2B, Local Voice Models, Android
• The Build: A secure automation system that monitors WhatsApp, manages calendars, and handles call pickups locally. Built because automating personal tasks shouldn't require trading your private data to cloud providers.

• Domain: Infrastructure & Simulation
• Tech: Python, Computer Vision, Data Processing
• The Build: An end-to-end simulation and watch system (rail-sentry-watch / simulation-rails) to model railway metrics and detect track anomalies. A proof-of-concept for replacing manual, error-prone railway inspections with automated data analysis.

• Domain: Mobile & Web-to-Native Bridging
• Tech: Capacitor, React, Bun, Android Native
• The Build: Designed to eliminate the performance gap in hybrid apps. This high-performance client was natively verified for zero-bug deployments, proving that web technologies can achieve native-level execution when properly architected.

• Domain: Education & Applied AI
• Tech: Advanced AI Engine, Web Stack
• The Build: A highly specialized testing environment focusing on AI-driven insights for competitive exams. Combines advanced algorithmic test generation with real-time analytics to adapt to student performance dynamically.

• Domain: Healthcare & Edge Computing
• Tech: TensorFlow Lite (TFLite), Edge ML
• The Build: A finalized TFLite machine learning model designed to detect and manage Body-Focused Repetitive Behaviors (BFRB) on edge devices. Focuses on low-latency, private, on-device health monitoring without cloud latency.

• Domain: Backend Intelligence
• Tech: Python, API Architecture, ML Frameworks
• The Build: The standalone, production-ready AI brain (ai-engine-adv-production) used across various platforms (like Rankak and JEE Smart AI). Centralizes machine learning models into a highly available, robust microservice.

• Domain: Backend Architecture
• Tech: SQL, Database Design
• The Build: An architecture-focused repository aimed at structuring scalable, relational data systems. Ensures data integrity and fast query times for high-traffic platforms, emphasizing that AI is useless without a solid data foundation.

• Domain: UI/UX & Recommendation Systems
• Tech: Figma, Frontend Frameworks, ML Recommendations
• The Build: Combines clean, modern interface design (codsoft-ui-ux) with an AI-powered book recommendation engine. Proves that I don't just build the backend logic—I also construct the human-computer interfaces required to make those systems usable.