AWS Jobs in Manufacturing: What the Role Involves

# Store a sensor reading in S3 (needs AWS credentials configured)
# pip install boto3
import json
from datetime import datetime, timezone
import boto3

s3 = boto3.client("s3")
now = datetime.now(timezone.utc)

reading = {
    "plant": "plant-01",
    "line": "line-3",
    "machine": "press-07",
    "temperature_c": 74.5,
    "vibration_mm_s": 3.1,
    "timestamp": now.isoformat(),
}

key = f"sensor-data/plant-01/line-3/{now:%Y/%m/%d}/{now:%H%M%S}-press-07.json"
s3.put_object(
    Bucket="example-factory-data-bucket",   # replace with your bucket
    Key=key,
    Body=json.dumps(reading),
    ContentType="application/json",
)
print("stored", key)

S3 key layout for factory data

sensor-data/<plant>/<line>/<YYYY>/<MM>/<DD>/<time>-<machine>.json

Benefits:
- Easy to find one plant, line or day
- Query tools can scan only the days needed
- Lifecycle rules can move old data to cheaper storage
- Access can be limited per plant if required

Plan the layout before data starts arriving -
changing it later is painful.

# Everyday AWS CLI tasks (credentials configured with: aws configure)

# List today's files for one line
aws s3 ls s3://example-factory-data-bucket/sensor-data/plant-01/line-3/ --recursive

# Copy a day of data locally for investigation
aws s3 cp s3://example-factory-data-bucket/sensor-data/plant-01/line-3/2026/11/05/ ./day-data/ --recursive

# Publish a custom metric for a dashboard or alarm
aws cloudwatch put-metric-data --namespace Factory/Line3 \
  --metric-name PressTemperature --value 74.5 --unit None

# Check who you are logged in as
aws sts get-caller-identity

Cloud on-call handover (sample)

Date: ____   Engineer: ____

Systems status:
- Data ingestion line 1-4: normal
- Line 3 data delayed 20 min at 02:00, recovered
- Maintenance app: normal

Alarms raised: 1 (Line 3 ingestion delay)
Root cause: network outage at plant gateway
Follow-up: ask plant IT for gateway logs
Open risks: disk space on reporting server at 80%

-- Amazon Athena: average temperature per machine per day
-- (assumes a table 'sensor_readings' defined over the S3 JSON files)
SELECT machine,
       date(from_iso8601_timestamp(timestamp)) AS reading_date,
       round(avg(temperature_c), 1)            AS avg_temp_c,
       max(temperature_c)                      AS max_temp_c
FROM sensor_readings
WHERE plant = 'plant-01'
GROUP BY machine, date(from_iso8601_timestamp(timestamp))
HAVING max(temperature_c) > 80
ORDER BY reading_date DESC, max_temp_c DESC;

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "WriteLine3SensorDataOnly",
      "Effect": "Allow",
      "Action": "s3:PutObject",
      "Resource": "arn:aws:s3:::example-factory-data-bucket/sensor-data/plant-01/line-3/*"
    }
  ]
}

This lets an uploader write only to its own line's folder.
It cannot read, delete or touch other plants' data.

Architecture Brief - Machine Data Platform

Data sources: _______________ (machines, sensors)
Frequency: every ___ seconds
Ingestion: AWS IoT Core / API / batch upload
Storage: S3 bucket, key layout: _______________
Processing: Lambda / Glue
Analytics: Athena / dashboards
Security: IAM roles, device certificates, encryption
Monitoring: CloudWatch alarms on ___
Data retention: ___ months, then cheaper storage

Runbook - Sensor Data Not Arriving

1. Check CloudWatch alarm details and start time
2. Is the problem one machine, one line or one plant?
3. Ask plant IT: is the site gateway online?
4. Check ingestion errors in logs
5. Confirm IAM permissions were not changed recently
6. When fixed, confirm backlog data arrived
7. Write a short incident note with root cause

Quarterly Access Review

User / role            Access            Still needed?  Action
line3-uploader         PutObject line-3  Y              keep
contractor-2025        S3 read all       N              remove
analyst-team           Athena read       Y              keep
old-test-role          Admin             N              remove