Skip to main content

​
Overview

run_feature_worker is the second step of the pipeline. It triggers the job-feature-worker pipeline job, which reads raw OHLCV data from blob storage, computes technical indicators via TA-Lib, and writes the enriched feature dataset back to storage. The tool blocks until the job completes and returns the output blob URL.

​
Parameters

​
input_url
string
required
Blob storage URL pointing to a data_extractor_*.json file.This is the output_url returned by run_data_extraction.
​
config
object
required
Feature engineering configuration.

​
Returns

{
  "status": "Succeeded",
  "output_url": "https://stmcpfabricdev.blob.core.windows.net/data/feature_engine_20260307_125103.json",
  "output_name": "feature_engine_20260307_125103.json",
  "execution_name": "job-feature-worker-abc123xyz"
}
FieldDescription
statusJob terminal status (Succeeded)
output_urlFull HTTPS URL to the output blob — pass to run_ml_job or run_dl_job
output_nameBlob filename
execution_nameJob execution ID for audit/debugging

​
Example

{
  "input_url": "https://stmcpfabricdev.blob.core.windows.net/data/data_extractor_20260307_124937.json",
  "config": {
    "Features params": {
      "bb_lines": { "status": false, "windows": [20] },
      "moving_averages": { "status": true, "windows": [5, 15, 22] },
      "rsi_indicators": { "status": true, "windows": [5, 15] },
      "macd_indicators": { "status": true, "windows": [12] },
      "stochastic_indicators": { "status": false, "windows": [14] },
      "atr_indicators": { "status": false, "windows": [14] },
      "adx_indicators": { "status": false, "windows": [14] },
      "cci_indicators": { "status": false, "windows": [14] },
      "roc_indicators": { "status": false, "windows": [10] },
      "obv_indicator": false,
      "candlestick_patterns": { "status": false }
    }
  }
}

​
Resources

ResourceValue
Container Apps Jobjob-feature-worker
Container namefeature-worker
Env vars injectedINPUT_URL, CONFIG
Output blob prefixfeature_engine_
Timeout600 seconds

​
Next Step

Pass output_url (along with the original data_extractor_ URL) to either: