Home/Use Cases/Long-Running Jobs

Run Jobs for Hours or Days — No Timeouts

Data processing, scraping, ML training, batch jobs — execute workloads that run for as long as they need. Full state persistence, real-time monitoring, zero server management.

No credit card requiredUp to 30 days runtimePay per second

Tired of Timeout Limits?

Serverless platforms are great — until your job needs more than 15 minutes. Then you're stuck with painful workarounds:

  • Breaking jobs into tiny chunks and managing state yourself
  • Spinning up EC2 instances and dealing with DevOps overhead
  • Paying for idle time while waiting for jobs to complete
  • Complex orchestration with Step Functions or Airflow

With Hopx

  • Start a sandbox and let your job run — for minutes, hours, or days
  • Full filesystem for checkpoints and intermediate results
  • Stream logs and monitor progress in real-time
  • Pay only for actual compute time — per second billing

Perfect For Any Long-Running Workload

Data Processing

ETL pipelines, data transformations, batch analytics

Web Scraping

Large-scale crawling, data extraction, monitoring

ML Training

Model training, hyperparameter tuning, experiments

Video Processing

Transcoding, analysis, thumbnail generation

Report Generation

Complex reports, PDF generation, data exports

Database Migrations

Schema changes, data backfills, cleanup jobs

Built for Long-Running Workloads

No Timeout Limits

Run jobs for hours, days, or weeks. No 15-minute Lambda limits, no forced shutdowns, no arbitrary timeouts.

Full State Persistence

Filesystem persists across the job lifetime. Write checkpoints, resume from failures, store intermediate results.

Real-Time Monitoring

Stream logs via WebSocket, track CPU/memory usage, get health metrics — all through simple APIs.

No Server Management

No EC2 instances to provision, no containers to orchestrate. Just start your job and let it run.

Hopx vs. Alternatives

FeatureHopxAWS LambdaFargate
Max RuntimeUnlimited (days/weeks)15 minutesHours (with config)
Cold Start~100ms100ms-10s30s-2min
State PersistenceFull filesystemNoneEFS (extra cost)
Pricing ModelPer-secondPer-request + durationPer-hour
Setup ComplexityZeroMediumHigh

Simple API for Complex Jobs

Start a background job, monitor progress, download results. No infrastructure code, no DevOps complexity.

Background Execution

Start jobs that run independently while you do other work

Live Monitoring

Stream logs and check status without interrupting the job

Checkpoint & Resume

Save progress to filesystem, resume from failures

long_job.py
1from hopx_ai import Sandbox
2
3# Create sandbox for long-running job
4sandbox = Sandbox.create(
5    template="code-interpreter",
6    timeout=None  # No timeout - run indefinitely
7)
8
9# Start background job
10process = sandbox.run_code_background("""
11import time
12
13# Your long-running job
14for batch in range(1000):
15    process_batch(batch)
16    
17    # Checkpoint progress
18    with open('/workspace/checkpoint.json', 'w') as f:
19        json.dump({'batch': batch}, f)
20    
21    print(f"Processed batch {batch}/1000")
22""")
23
24# Monitor progress
25while True:
26    status = sandbox.get_process_status(process.id)
27    logs = sandbox.get_logs()
28    print(logs[-10:])  # Last 10 log lines
29    
30    if status.completed:
31        break
32    time.sleep(60)
33
34# Download results
35results = sandbox.files.read("/workspace/results.csv")
36sandbox.kill()

Stop Fighting Timeout Limits

Get $200 in free credits. Run your first long job in minutes. No credit card, no DevOps setup.