MercorResearch and academiaReviewed
LLM Research Scientist (Pre-training & Post-Training)
- Platform
- Mercor
- Fit category
- Research and academia
- Remote/location
- Fully remote; current work-authorization requirements apply
- Pay
- Firm hourly pay: $100 - $120 / hour
- Eligibility
- Remote, work authorization mentioned
- Difficulty
- Highly selective
- Last reviewed
- Jul 10, 2026
What this role involves
Mercor is recruiting experienced machine-learning researchers for project-based work on open-ended LLM research problems. The role spans transformer pre-training, web-scale corpus construction, data and compute allocation, supervised and preference-based post-training, reward modeling, alignment, evaluation, and diagnosis of training instability.
Location / eligibility
Remote, work authorization mentioned. Listed location: Fully remote; current work-authorization requirements apply. Some opportunities are country-, credential-, tax-, or work-authorization-specific. Check the current Mercor page before applying.
Requirements
- At least three years of machine-learning research experience; PhD research counts
- Hands-on experience training transformer language models or improving them through fine-tuning or post-training
- Strong experience with PyTorch, JAX, TensorFlow, or a similar ML framework
- Expertise in pre-training, corpus construction, DPO/RLHF/RLAIF, reward modeling, alignment, or rigorous LLM evaluation
- A degree from a top-100 university, experience at a major AI or technology company, or an equivalent publication/open-source research track record
Best for
- ML researchers with end-to-end transformer training experience
- PhDs or industry researchers working on LLM pre-training, post-training, or evaluation
- Researchers with strong publications, open-source contributions, or major AI-company experience
Skip this if...
- You do not have at least three years of machine-learning research experience
- Your experience is limited to API use without model training, data, or evaluation research
- You require H-1B or STEM OPT support and the live listing still says those candidates are not supported
- You need guaranteed acceptance, work, pay, hours, or project duration
Application tips
- Verify the current Mercor page before applying because availability and project duration can change.
- Prepare concise examples of pre-training, post-training, corpus construction, evaluation, or training-debugging work.
- Use public papers, repositories, or non-confidential project summaries to demonstrate research impact.
Mercor listing: Open this opportunity on Mercor