pawsMatch

Inspiration If you’ve ever volunteered at an animal rescue, you know it’s organized chaos. Shelters are chronically understaffed, and they rely heavily on whiteboards, scattered spreadsheets, and memory to pair animals with adopters. The heartbreak isn't just that there are too many animals; it's that great applicants often fall through the cracks, and pets stay in kennels far longer than necessary just because matching them up takes so much manual effort. Furthermore, in diverse communities, language barriers can prevent incredibly loving families from ever being considered. We wanted to build something that felt less like a cold database and more like an intelligent, empathetic assistant for shelter coordinators.

What it does pawsMatch is an intelligent shelter operating system designed to bring harmony to the adoption process. At its core, it features a one-click compatibility engine that instantly evaluates every available pet against every active applicant. Instead of just blind guessing, it uses a smart algorithm to automatically veto unsafe matches (like placing a resource-guarding dog in a bustling, inexperienced household) and bubbles the best fits to the top. It even generates a clear, human-readable explanation of why a specific match works perfectly.

Beyond matching, pawsMatch includes an AI-powered agent that helps staff query the database, manage capacity escalations (flagging when the shelter is getting too full), and instantly translate intake messages and applications into regional Indian languages to break down accessibility barriers for non-English-speaking adopters.

How we built it We knew shelter workers spend hours staring at screens, so we wanted to build a modern, lightning-fast, and beautiful experience.

Frontend: We used React 19, TypeScript, Vite, and TailwindCSS to build a stunning, fluid dashboard. We leaned heavily into motion for micro-animations to make the interface feel alive and premium, moving away from the clunky feel of traditional enterprise software. Backend: We built a fast, robust REST API using Python and FastAPI, backed by an asynchronous SQLite database (aiosqlite) for lightweight, reliable data persistence. AI & Logic: We implemented a hybrid approach. The core scoring engine is deterministic and rule-based, ensuring absolute safety and transparency. We then layered in an agentic AI framework (utilizing Sarvam AI and Strands) to handle the natural language queries, generate human-readable match explanations, and run the real-time translation workflows. Challenges we ran into One of our biggest hurdles was figuring out where to draw the line between AI and traditional logic. Initially, we thought about just handing all the matching over to a Large Language Model. But we quickly realized that in animal welfare, shelter coordinators need absolute transparency. If an AI hallucinated a "perfect match" for a highly reactive dog, it could result in a dangerous situation. We had to pivot our architecture mid-hackathon to a hybrid model: strict, rule-based algorithmic scoring for safety and ranking, which is then augmented by AI to explain the reasoning, translate languages, and help coordinators query the data.

Accomplishments that we're proud of We are incredibly proud of the user experience. Shelter software is notoriously dated and frustrating to use. We managed to build a platform that looks and feels like a premium consumer app while handling complex logistical operations under the hood.

We are also exceptionally proud of the regional language translation feature. Knowing that our tool can make pet adoption accessible to families who don't speak English feels like a massive win for both the animals and the community.

What we learned We learned so much about the deep nuances of animal welfare. We realized that "compatibility" isn't just a simple checklist of "dog + yard = good." It requires evaluating human experience levels, existing household dynamics, and subtle behavioral traits. On the technical side, we leveled up our ability to build agentic AI frameworks that can safely interact with a backend database without going off the rails.

What's next for pawsMatch We want to expand the AI agent's capabilities so it can autonomously draft personalized outreach emails and text messages to top-tier applicants. We're also exploring a live WhatsApp bot integration, allowing applicants to get status updates and submit their initial forms directly from their phones in their native language. Ultimately, our goal is to pilot pawsMatch with a local rescue to see our code start saving real lives.