Back to projects
elixir
ai
Distributed AI Streaming Platform
Multi-region AI response streaming with Erlang clustering, GraphQL subscriptions, and OpenTelemetry.
Problem
AI assistants needed low-latency streaming across regions without single-node bottlenecks or blind spots in observability.
Role
Senior backend architect
Stack
Elixir, Erlang clustering, Phoenix PubSub, GraphQL, OpenTelemetry, AWS, Kubernetes
Result
Delivered a fault-tolerant streaming layer spanning multiple AWS regions with subscription-based delivery and production-grade tracing.