Machine Learning Scientist - AI Trainer

DataAnnotation; Oklahoma; Remoto; USD 124.000 - 207.000; Tempo pieno

Vantaggi offerti da questo lavoro

Flexible, remote work

Descrizione del lavoro

DataAnnotation is seeking a Machine Learning Engineer to stress-test how models reason about ML systems, including training dynamics, evaluation design, data pipelines, and deployment trade-offs. The role focuses on crafting prompts that reveal reasoning gaps and validating model outputs against real-world expectations.

You’ll review outputs, identify subtle errors, and write correct, principled solutions grounded in practical experience.

Competenze

• Hands-on experience training, evaluating, or deploying models professionally or in serious personal work.
• Comfort with the modern ML stack; most tasks assume Python and PyTorch or JAX.
• Clear written English: your explanations are the training signal.
• No degree required. We care about what you can do, not where you learned it.

Mansioni

• Write prompts that probe how models reason about training, evaluation, debugging, and productionizing ML systems.
• Review AI output for subtle errors: leaky evaluations, wrong loss formulations, and misdiagnosed training failures.
• Write the correct solution when the model falls short, grounded in real practitioner experience.

Conoscenze

ML engineering

Python

PyTorch

JAX

Model evaluation

Productionizing ML

Strumenti

PyTorch

JAX

Descrizione del lavoro

Models are surprisingly bad at reasoning about themselves: training dynamics, evaluation design, data pipelines, and deployment trade-offs. Plausible-sounding ML advice is often subtly wrong.

As a Machine Learning Engineer you'll stress-test how models reason about ML systems and write the answers a strong practitioner would give, shaping how the next generation handles your field.

What you’ll actually do

• Write prompts that probe how models reason about training, evaluation, debugging, and productionizing ML systems.
• Review AI output for subtle errors: leaky evaluations, wrong loss formulations, and misdiagnosed training failures.
• Write the correct solution when the model falls short, grounded in real practitioner experience.

What we look for

• Hands-on experience training, evaluating, or deploying models professionally or in serious personal work.
• Comfort with the modern ML stack; most tasks assume Python and PyTorch or JAX.
• Clear written English: your explanations are the training signal.
• No degree required. We care about what you can do, not where you learned it.

Compensation

Up to $150+/hr depending on task difficulty and specialization. Many contributors add $100k–$200k a year; some make it their full-time income.

About DataAnnotation

DataAnnotation is where 1M+ experts train the world’s leading AI models. Hundreds of millions paid to contributors to date, and the average contributor stays 5+ years. Flexible, remote, and always project-available.