Student developer exploring AI, cybersecurity, and product building. Open source contributor and hackathon enthusiast.

projects

github

youtube

medium

Tirup Mehta • Peerlist

<h3>About Liquid Ascii</h3>Liquid Ascii is a free, open-source React component that renders a real-time FLIP fluid simulation as ASCII text on a canvas. It offers a unique visual experience where fluid dynamics are represented by characters, creating an engaging and dynamic display. The component is designed with zero dependencies and is MIT licensed, allowing for unrestricted use and modification.<h3>Key Features:</h3><ul><li>Real-time FLIP Fluid Simulation: Experience a dynamic fluid simulation rendered entirely as ASCII characters.</li><li>Zero Dependencies: Easily integrate into your projects without worrying about additional library conflicts.</li><li>MIT Licensed: Free for commercial and personal use, with the freedom to copy, modify, and distribute.</li><li>Interactive Playground: Live playground available to experiment with different settings and see the results instantly.</li><li>Customizable Props: Control various aspects of the simulation, including speed, cell size, gravity, fill height, cursor interaction, and auto-wave effects.</li><li>Font Flexibility: Supports any CSS font face, allowing you to customize the glyphs used in the simulation.</li><li>Responsive Design: Adapts to light and dark modes automatically and respects prefers-reduced-motion settings.</li></ul><h3>Installation</h3>Install Liquid Ascii easily into your Next.js or React project using the following command:npx shadcn@latest add https://tirup.in/components/r/liquid-ascii.jsonThen, render the component within any sized container in your application.<h3>Technical Details</h3>The simulation involves up to 20,000 particles on a staggered MAC grid. Each frame calculates gravity, splats velocities to grid faces, solves for pressure over 30 iterations to maintain volume, and blends the results back to particles using a PIC/FLIP method. Density is binned per cell and rendered using glyphs, with character density varying based on surface, spray, and depth.

Liquid Ascii

<h3>About ModelRegistry</h3>ModelRegistry is a comprehensive, open-source technical ledger that meticulously tracks and indexes primary foundation model architectures and specialized research checkpoints from leading AI laboratories worldwide. It serves as a central hub for the latest advancements in artificial intelligence, providing detailed insights into the cutting-edge developments from organizations such as OpenAI, Anthropic, Google DeepMind, Meta AI, xAI, Qwen, Mistral, Tencent, Z.ai, and many others.<h3>Key Features and Data Points:</h3><ul><li>Comprehensive Model Tracking: Monitors flagship models and research checkpoints from a wide array of premier AI research labs.</li><li>Detailed Specifications: Includes source-linked specifications, context limits, pricing information, release dates, and lab-published benchmark metrics for active frontier AI foundation models.</li><li>Open and Accessible: Promotes transparency and accessibility within the AI research community.</li><li>Community Contributions: Encourages contributions from the open-source community to maintain and expand the registry.</li><li>Data Formats: Provides data in multiple formats including JSON, RSS, and plain text for easy integration and analysis.</li><li>Up-to-date Information: Regularly updated to reflect the latest model releases and advancements, with a focus on current and future trends in AI development.</li></ul>ModelRegistry aims to be the definitive resource for understanding the rapidly evolving landscape of frontier AI models, empowering researchers, developers, and enthusiasts with the information they need to stay at the forefront of artificial intelligence.

ModelRegistry

<h3>About Me</h3>I build interfaces where visual clarity meets system performance. Working at the intersection of design and front-end engineering, I focus on creating fast, robust web applications that feel natural to use.Driven by a strong curiosity for cryptography and systems security, I design software with data safety built into its foundation, bridging the gap between secure backend architecture and polished visual flows.<h3>Contributions and Engagement</h3><ul><li>I write about interface design and web security on my blogs.</li><li>I share active experiments on GitHub.</li><li>I am always open to fresh ideas and connections.</li></ul>

Tirup Mehta | Designer & Developer

Trace Guard is an advanced behavioral security engine and open-source NPM package built to detect and block bots, scrapers, automated browsers, and AI-driven agentic traffic in real time. Instead of relying on traditional CAPTCHA systems or static fingerprinting, it silently intercepts HTTP/HTTPS requests, injects lightweight behavioral telemetry scripts, and analyzes user interaction patterns to differentiate humans from automated systems. The project focuses on modern attack vectors including headless browsers, automation frameworks, and Vision-Language-Model-driven agents capable of mimicking human behavior. Trace Guard was designed with zero-friction integration in mind — developers can secure applications by adding a single middleware layer to existing Node.js infrastructures. Key highlights include behavioral analysis, stealth telemetry injection, real-time request interception, bot mitigation, lightweight performance overhead, and seamless developer experience through NPM distribution. The project combines cybersecurity research, browser behavior analysis, and modern web infrastructure engineering into a production-ready security package aimed at protecting modern applications from evolving automated threats.

Trace Guard

Vectorize is a contextual AI infrastructure built for modern companies. It empowers small and mid-sized teams to deploy AI solutions with embedded business intelligence. By registering organizational metadata, companies enable domain-specific, context-aware responses. Persistent memory allows conversations to retain relevance across multiple interactions. The system outputs fully white-labeled results, free of third-party references or model attribution. Real-time API key validation and secure authentication flows streamline onboarding. Middleware handles internal and external routing with consistent logic. An integrated dashboard provides visibility into usage, session history, and profile configuration. SDKs in multiple languages accelerate development across diverse stacks. Vectorize bridges enterprise-grade AI capabilities with practical, developer-first access.