AI-Powered Incident Prevention Platform for Urbint

At Variant Systems, we collaborated with Urbint, a leader in AI-driven risk prevention, to build a platform that helps utility companies prevent incidents before they happen.

🧠 The Challenge

Urbint needed to scale their incident prediction platform, which leverages AI to assess risk and prevent field incidents across critical infrastructure systems. Their team required hands-on technical expertise to accelerate delivery while maintaining high code quality.

🔧 Our Contribution

We worked closely with Urbint’s engineering team to:

• Enhance backend services using Python, focusing on data processing and model integration, improving performance of existing services and fixing scaling challenges
• Revitalized a legacy geospatial data service that lacked documentation and active maintainers, causing frequent crashes. We tackled the challenge by stabilizing the service, addressing critical failures, and creating comprehensive documentation. This effort not only made the service more maintainable and scalable but also laid the groundwork for a future full rewrite
• Worked extensively with GeoJSON and PostGIS to manage and process geospatial data, optimizing spatial queries and ensuring accurate and efficient data handling for Urbint’s risk analysis tools
• Developed a custom form builder application in React for Urbint’s multi-tenant platform. This tool enabled admins and customer success teams to easily create tailored forms for clients, streamlining the process of client-specific data collection and reducing manual work
• Maintained legacy frontend code written in functional TypeScript with a Haskell-inspired functional programming style using fp-ts. We modernized key parts of the frontend, refactored components, and ensured backward compatibility while introducing performance improvements
• Ensured tight feedback loops and code quality through embedded collaboration and agile delivery

Our developers operated as an extension of Urbint’s in-house team, bringing technical depth, architectural input, and delivery momentum.

🚀 Results

Through our collaboration, we successfully:

• Improved platform stability by addressing critical backend failures and optimizing geospatial data processing, ensuring the service could scale efficiently with Urbint’s growing user base.
• Reduced operational downtime and incidents by modernizing and maintaining the legacy systems, enabling Urbint to continue delivering value to their clients without major interruptions.
• Accelerated feature delivery by enabling admins to quickly create custom forms for clients through our form builder tool, reducing manual efforts and providing a more flexible solution for their multi-tenant platform.
• Enhanced user experience by improving frontend responsiveness and performance, ensuring users could interact with the platform more seamlessly across devices and use cases.
• Laid the foundation for future innovation by documenting key parts of the system and setting up an architecture that could evolve into a more robust, long-term solution.

🛠 Tech Stack

• Frontend: React, Zustand, TypeScript, fp-ts, TanStack Query
• Backend: Python, FastAPI, GraphQL, GeoJSON, Microservices
• Database: PostgreSQL (with PostGIS extensions for geospatial data)
• Infrastructure: Docker, Docker Compose, GitHub Actions, Various Cloud Platforms
• Practices: Agile, Test-Driven Development, Continuous Integration, Continuous Delivery