Modern scientific research is often fragmented. A single project might require researchers to jump between terminal windows for HPC job submission, cloud storage for data files, messy email threads for documentation, and external AI tools for analysis. Many researchers don't know how to use HPCs so being able to abstract the process of inputting data and documents and getting an output where it is simple enough to understand was the final goal. We were inspired to build a "unified research operating system"—a single glass pane where a wildfire scientist or a quantum physicist could manage their team, run heavy simulations, and have an AI instantly translate cryptic log files into professional, statistical takeaways. We developed this application using React and Tailwind CSS, integrated with the Gemini API for expert data interpretation and Recharts for real-time simulation tracking.One major challenge was making the "HPC Cluster" feel friendly. Most HPC interfaces are daunting command-line environments. We had to design an abstraction (the HPCPortal) that felt powerful enough for engineers but simple enough for stakeholders to track progress. One of the biggest takeaways was the importance of Contextual AI. We learned that an AI is only as good as the metadata you provide it. By feeding Gemini the specific project goals, team roles, and staged input files, we transformed a general LLM into a "domain-specific" research assistant that understands the difference between a drag coefficient in an EV wind tunnel and a qubit coherence state in a quantum lattice.
Built With
- gemini
- gemini-studio
- html
- typescript
Log in or sign up for Devpost to join the conversation.