Quick facts
- Type
- Contract
- Location
- Remote (US, Canada)
- Skills
- PyTorch, TensorFlow, Machine Learning
- Platform
- Workada — 31 open jobs
- Typical pay at Workada
- $23–130/hour · median $100 (9 jobs on our site)
- Payout
- Stripe (3 times a week)
- Apply
- Free to apply — never pay to get AI training work
What you'll do
- Build and train a model in PyTorch or TensorFlow on a public dataset
- Fine-tune a pretrained model from the Hugging Face Hub and evaluate it
- Debug training issues, tune hyperparameters, and track metrics
- Package a model for inference or push it to the Hugging Face Hub
- Export the code, notebook, or model card you produced
- Work at your normal professional pace and quality, the way you would for a real client or employer
Who we're looking for
- At least 2 years of hands-on machine learning work using one or more of these frameworks
- Your own working environment (a local GPU, Colab, or a free cloud tier is fine)
- A Mac (Apple Silicon or Intel) or Windows computer you can install software on
- Willingness to install our recording app, Bloom, and grant it Screen Recording and Accessibility permissions
- Ability to complete identity verification
- Use only sample data or data you own: no confidential employer or client information on screen
How it works
- Apply and select the tool you're strongest in.
- If you're a fit, complete onboarding: identity verification, Workada Timer setup, Bloom setup, and accepting the contributor terms.
- Open Bloom, start recording, and complete about an hour of real work in your ML stack.
- Submit the recording along with the file you produced.
What we offer
- Fully remote, contract work you schedule around your life
- Paid work that uses skills you already have
- A chance to shape how AI learns to use professional software