
1
1
India
3+ years of experience
I’m Ayush Tyagi, a forensic science student with a strong passion for technology and software development. I combine analytical thinking from forensic science with practical coding skills to approach problems in a structured and evidence-driven way. Alongside my academic background, I actively work on development projects and explore modern technologies, focusing on building efficient and impactful solutions. I’m particularly interested in areas where technology intersects with investigation, security, and real-world problem solving.

ClimateGuard is a full-stack AI climate risk intelligence solution designed to enable people and organizations to analyze and mitigate climate risks in real time.The platform provides weather data, earthquake data, wildfires tracking, and air pollution data in one interface for users to create a climate risk dashboard. By typing any city across the globe, the user can immediately get the total climate risk score, severity of climate issues in that city and AI-recommended advice based on user’s profile. Qwen 2.5 1.5B is used to run AI assistant locally with inference on AMD MI300X GPU through ROCm and Ollama. With all computations made locally, ClimateGuard does not need any remote APIs to produce AI-powered responses.The AMD MI300X equipped with 192GB HBM3 memory makes it a perfect device to run large language models fully precise. React and Vite are used for building the frontend, Node.js/Express serves as a backend with Autonomous agent scheduling which updates the risk information every 15 minutes. Multiple climate APIs are also used (OpenWeather API, NASA FIRMS API, and GDACS). Finally, ClimateGuard was deployed on AMD Developer Cloud through a droplet that contains GPU and supports ROCm for real-time use. This allows the application to
10 May 2026