✦ Frontier Systems & AI Alignment • 2026 Portfolio

Engineering Zero-Overhead Computer Use & Empirical AI Alignment

Architect of native high-speed Model Context Protocol (MCP) engines, deterministic anti-injection firewalls, and enterprise-scale academic intelligence operating systems.

🚀 Explore Frontier Projects ⚡ Live MCP Sandbox 👤 Architect Bio & Focus

150k+

Lines of Systems Code Built

< 0.1ms

Kernel Input Syscall Latency

80–90%

Vision Token Reduction (Visual Diff)

AI Safety Benchmarks Audited

Projects Showcase

Select any project to explore its full architectural breakdown, benchmarks, and source details.

⚡ Interactive MCP Tool Sandbox

Test real Model Context Protocol JSON-RPC dispatches against simulated native runtimes.

Native Rust & TS Runtimes

omnidesk-session://stdio:transport [v2.1.0]

Ping: 0.8ms

Dispatch Tool Call:

About Ashutosh Chikane

Undergraduate systems engineer and AI alignment researcher at the Department of AI & Machine Learning, Walchand College of Engineering (WCE), Sangli, India.

I specialize in architecting high-performance runtime bridges between frontier AI models and local operating systems. Rather than treating computer use as crude mouse-clicking, I design zero-overhead systems using native Win32 message pipelines, hardware-accelerated WinRT OCR, and deterministic security firewalls that keep users fully in control.

✉️ Email Me 🐙 GitHub 💼 LinkedIn

Core Systems Competencies

Systems & Runtime:

Rust Win32 API WinRT OCR COM Automation TypeScript Puppeteer / CDP

AI & Agentic Architectures:

Model Context Protocol (MCP) Constitutional AI (CAI/RLAIF) Set-of-Marks (SoM) Grounding Perceptual Visual Diffs HarmBench / WMDP Auditing

Frontend & Visualization:

HTML5 Canvas API (60 FPS) Modern Vanilla JS (ES6+) SvelteKit 2 Glassmorphic CSS KaTeX Typography

Engineering Zero-Overhead Computer Use & Empirical AI Alignment

Projects Showcase

⚡ Interactive MCP Tool Sandbox

About Ashutosh Chikane

Core Systems Competencies

⚡ Power User Shortcuts