# continuum-ai-worker

Computer-vision failure detection worker for the Continuum print farm
platform. Consumes camera frame batches pushed onto the Redis Stream
`stream:camera_frames` (by `continuum-backend`'s `src/queue/`), runs them
through an ONNX Runtime model, and fast-paths high-confidence failure alerts
back to Redis and Postgres.

## Pipeline

```
continuum-proxy --(frames)--> continuum-backend --(XADD)--> stream:camera_frames
                                                                    |
                                                          XREADGROUP (this worker)
                                                                    v
                                                        src/inference (ort session)
                                                                    v
                                                  src/reporter (Redis pub/sub + Postgres INSERT)
```

## Structure

```
src/
  main.rs         Consumer-group loop: XREADGROUP -> decode -> infer -> report -> XACK
  inference/       Thread-safe ONNX Runtime session wrapper + pre/post-processing
  reporter/        Alert fan-out: Redis pub/sub for realtime UI, Postgres for audit trail
```

## Getting started

```bash
cp .env.example .env
# place an ONNX failure-detection model at ./models/failure-detector.onnx
cargo run
```