GridShadow

Inspiration

Every year, storms leave thousands of people without electricity. While weather forecasts can predict severe conditions, utility operators still have to make difficult decisions under pressure—deciding which areas are most at risk, where to send repair crews, and how to restore power as quickly as possible.

I wanted to build an AI system that doesn't just predict bad weather—it helps people decide what to do before the storm arrives.

That's how *GridShadow * was born.

What it does

GridShadow is an AI-powered storm operations copilot that helps utilities and microgrids prepare for extreme weather.

Instead of overwhelming operators with weather data, GridShadow transforms forecasts into clear, actionable decisions.

The platform can:

Monitor live weather forecasts and severe weather alerts
Identify which areas are most likely to experience outages
Rank zones and infrastructure by operational risk
Prioritize hospitals, emergency shelters, water systems, and other critical infrastructure
Recommend where crews and equipment should be deployed before the storm arrives
Generate restoration plans after outages occur
Automatically create operational briefings and emergency reports

One of GridShadow's biggest strengths is personalization.

Every utility operates differently. By learning from uploaded operating procedures, restoration playbooks, previous incident reports, and internal documentation, GridShadow adapts its recommendations to each organization instead of providing generic advice.

How I built it

GridShadow combines several AI-powered components into one workflow.

Frontend

React
Tailwind CSS

Backend

Flask
Python

AI

Gemini API for intelligent planning, summarization, and operational recommendations

Authentication

Auth0

Data Processing

Weather forecast APIs
Risk scoring engine
Document-based memory system
Retrieval-Augmented Generation (RAG) for personalized recommendations

The overall workflow is:

Weather Forecast → Risk Analysis → AI Decision Support → Personalized Restoration Plan → Operations Brief

Challenges I ran into

One of the biggest challenges was moving beyond a simple weather dashboard.

Many existing solutions stop after displaying forecasts or alerts. I wanted GridShadow to answer the next question:

"What should operators actually do now?"

Designing an AI system that converts weather information into practical operational decisions while considering critical infrastructure, restoration priorities, and organization-specific procedures required multiple iterations.

Balancing customer restoration speed with protecting essential services was another interesting challenge.

Accomplishments that I am proud of

I am proud that GridShadow goes beyond predicting storms—it helps people respond to them.

Some highlights include:

Personalized AI recommendations instead of one-size-fits-all advice
Automatic prioritization of critical infrastructure
AI-generated operations and emergency response briefs
End-to-end workflow from forecast to restoration planning
A practical solution with real-world impact during extreme weather events

What I learned

During this project I learned how difficult emergency decision-making becomes when information comes from many different sources.

I also explored how large language models can move beyond chatbots and become operational decision-support systems that help professionals work faster and more confidently.