Asynchronous Batch Forecast

The OpenAPI specification defines task submission and status polling schemas for asynchronous batch jobs. It is designed for massive datasets referenced via pre-signed CSV or Apache Parquet object storage URLs.

1. Submission and Polling Lifecycle

A successful POST /batch/forecast returns HTTP 202 with a job_id and initial status per the specification. Clients subsequently poll GET /batch/jobs/{job_id} to retrieve progress and pre-signed output download URLs.

batch-contract.sh

curl -X POST "https://api.sequenxai.com/v1/batch/forecast" \
-H "Authorization: Bearer $SEQUENX_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: your-idempotency-key" \
-d '{
  "input_file_url": "https://storage.example.com/presigned/timeseries.parquet",
  "format": "parquet",
  "h": 24,
  "freq": "1h",
  "level": [80, 90]
}'

Contract note: OpenAPI also documents …/cn/v1/…; that regional prefix is not edge-routed yet. Use /v1 once the endpoint exists.

2. Production Evaluation Considerations

Dataset size thresholds, queue priorities, turnaround latency, polling intervals, and object storage compatibility will be finalized and documented in tandem with production infrastructure deployment.