Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
LLM Failover Chains and Redis
A technical walkthrough of building multi-LLM failover logic using provider selection, circuit breakers with Redis, and cost-aware fallbacks for real-world API resilience.
A technical walkthrough of building a multi-LLM orchestration layer for my job application automation tool. I’ll show the actual code that handles:
Provider selection logic with automatic failover (Claude → OpenAI → Gemini)
Circuit breaker pattern for serverless - how to persist failure state across Vercel cold starts using Redis
Model fallback when newer models aren’t accessible (handling the “Claude 4.5 dropped but my API key doesn’t have access” problem)
Cost optimization through provider ordering - no fancy algorithms, just “try the cheap provider first for cheap tasks”
Live demo: triggering an AI generation and watching the provider selection happen in real-time via DevTools.