"Model Mesh team on AI Infra Summit Hackathon Hackathon"
Team Idea
We're building ModelMesh, a smart routing layer that sits between your app and multiple LLM providers — automatically picking the right model tier for each request based on complexity, latency needs, and cost, with automatic fallback when a model is unavailable or rate-limited. Instead of hardcoding one model for everything, ModelMesh lets you get near-top-tier answer quality at a fraction of the cost by only escalating to expensive models when the request actually needs it. We're looking for builders interested in LLM infrastructure, backend systems, and evaluation/benchmarking to help build the routing logic and the cost/quality comparison harness that proves it works.
