$70–250/hr · Mercor · Part time
Machine learning engineer contributing specialized expertise to AI research projects through model development, training, and evaluation work.
What you would do
- Train and evaluate machine learning models for research-focused projects
- Design realistic task scenarios grounded in actual domain workflows
- Evaluate AI-generated outputs and provide actionable feedback for improvement
- Contribute technical insights to advance frontier AI research goals
Who they want
- Professional experience developing machine learning models and deploying them to production
- Strong command of Python and ML frameworks including PyTorch or TensorFlow
- Practical knowledge of model deployment, MLOps, and production considerations
- Ability to communicate clearly and work independently in remote environments
- Willingness to commit 15-30 hours weekly to flexible project work
Main skills
What the interview asks about
1.Model development iteration cycles
Understanding how to experiment systematically and evaluate improvements is core to advancing model capabilities.
For example: “You're training a classification model that achieves 85 percent accuracy but your domain knowledge suggests it should reach 95 percent. Describe the diagnostic steps you'd take to identify what's limiting performance.”
2.Framework selection and implementation
Choosing between PyTorch, TensorFlow, or other frameworks involves tradeoffs in flexibility, performance, and team standards.
For example: “You need to implement a model with custom loss functions and dynamic computation graphs. Explain why your framework choice suits these requirements.”
3.Production deployment considerations
Models that work in notebooks often fail in production due to latency, resource, or versioning issues.
For example: “Your research model requires 2 GB of memory and 500ms inference time. Describe the deployment challenges and optimization approaches you'd consider.”
4.Domain expertise in model evaluation
Metrics alone don't capture whether a model actually solves real problems in your domain.
For example: “An AI model makes accurate predictions but violates key domain constraints in how it reasons. Explain how you'd communicate this feedback to the research team.”
5.Remote collaboration and async work
Distributed teams require clear communication and ability to work without real-time guidance.
For example: “You discover an issue with task specifications that affects model training, but your project lead is unavailable. Describe how you'd document and escalate this problem.”
A task you may get
Build a simple machine learning model using your preferred framework, train it on sample data, evaluate results, and document findings and recommendations.
How to prepare
- Review fundamentals of your preferred ML framework and recent API changes
- Practice end-to-end model development from data preprocessing to evaluation
- Study how to communicate model limitations and domain-specific issues clearly
- Explore MLOps considerations like model versioning and deployment pipelines
The facts
- Pay
- $70–250/hr
- Hours
- Part time
- Where
- Remote
- Field
- Software Engineering
- Role type
- Talent network
- Posted
- 2/27/2026
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.