Data scientists and ML engineers occupy a rare position in the AI training market: you're not just evaluating a model's output, you understand how the model was built. Labs pay a premium for that — specialized ML roles regularly list at $150–$200/hr, among the highest rates on any AI training platform. And there's an unusual amount of it open right now, because frontier labs are pushing into narrow research areas that need genuine domain-plus-ML expertise.
Here's what the work is, the specialized roles open today, and how to position an ML background.
Two tiers of ML work
- Generalist ML / data science evaluation — judge model reasoning on statistics, feature engineering, model choice, and data-pipeline questions. $60–$120/hr. Open to most working data scientists.
- Specialized frontier-research roles — deep, narrow areas where the lab needs someone who has actually worked in the subfield. $150–$200/hr. This is where ML backgrounds shine.
Specialized ML roles open now
AfterQuery currently lists an unusually deep bench of research-tier ML positions — each a distinct role at $150–$200/hr:
- Machine Learning / Data Science Expert · Data Engineering Expert
- Diffusion Models · Image Generation · World Models
- Mechanistic Interpretability (LLMs) · Chip Design ML
- Proteomics · Bioinformatics · Computational Materials Science
- Network Science · Neuromorphic Computing · Robot Transfer Learning (PhD)
- Atmospheric / Climate / Geologic Modeling · Finance NLP
Browse them on our AfterQuery listings page or filter the whole board by ML / data science.
What the work actually involves
- Reasoning & solution evaluation — score whether model output reflects correct ML methodology, math, and domain assumptions.
- Reference-answer generation — write the gold-standard solution the model trains against (highest-paid).
- Red-teaming & edge cases — find where the model confidently gets the science or the stats wrong.
Which platforms hire ML talent
- AfterQuery — the deepest specialized-ML catalog (the roles above); apply directly to your subfield.
- Turing — engineering-heavy, strong for ML/infra engineers; see the micro1 vs Turing comparison.
- Mercor — engine-matched frontier-lab work for senior ML backgrounds, $150–$200+/hr.
How to position an ML résumé
- Name your subfield, not just "ML." "Diffusion models, 3 yrs at [lab]" routes you to the $200/hr role; "machine learning" routes you to the $80/hr one.
- List frameworks, papers, and shipped models. PyTorch/JAX, publications, production systems — concrete, verifiable.
- PhD/research signal helps at the top tier. Many specialized roles explicitly want research depth — see our PhD guide and frontier ML research jobs.
The AI-led interview rewards specifics — tactics in our interview guide.
Get started
Apply to your subfield on the AfterQuery page, add Mercor and Turing for engine-matched and engineering work, and read the complete AI training jobs guide for the full landscape. For pay context, see how much AI training jobs pay.