Machine Learning Expert

$150–200/hr

rate as listed

Workload: ~15h/week
Engagement: Project-based
Location: Worldwide
Field: STEM and Coding

Develop difficult machine learning engineering task ideas for a long-horizon research project, with an emphasis on inference, evaluation and systems problems. You must currently or recently have worked in an ML research role and have personally trained, fine-tuned or optimized models using PyTorch or JAX.

The work

You will propose machine learning engineering tasks intended to challenge frontier models over multiple attempts. Day to day, this means defining a technically precise task, explaining why it is difficult, and anticipating where a capable model will fail. The target is demanding: a strong task should succeed in no more than 2 of 8 attempts. The work leans toward inference, evaluation and systems rather than heavy model training, with roughly half of tasks running on GPUs and half on CPUs. You mainly contribute task plans and are not expected to build or review the completed tasks.

What the screen assesses

The platform screens for a current or recent role as an ML research engineer, research scientist or applied scientist, plus direct experience personally training, fine-tuning or optimizing models in PyTorch or JAX. Expect follow-up on at least one specialist area: LLM pretraining, alignment or evals; GPU inference and performance using vLLM, Triton, CUDA or profiling; reinforcement learning; or scientific ML in weather and climate, physics or biology. The screen is also likely to test whether you can distinguish a genuinely difficult, objectively evaluable task from one that is merely underspecified or laborious.

Practical details

Pay band: $150–200/hr

The listing does not specify remote status, asynchronous working arrangements, contract length or typical weekly hours. Candidates should be ready to state realistic availability and clarify scheduling expectations with Mercor. Benchmark-writing experience, a PhD in ML, CS or physics, an MSc with strong research output, publications and strong prior Mercor contract performance are preferences rather than stated requirements.

What they look for

Current or recent experience as an ML research engineer, research scientist or applied scientist
Hands-on model training, fine-tuning or optimization in PyTorch or JAX
Deep expertise in LLMs, GPU inference and performance, reinforcement learning or scientific ML
Ability to design difficult, objectively evaluable machine learning engineering tasks
Ability to explain concrete frontier-model failure modes
Inference, evaluation or ML systems judgment

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: This verifies that your work fits the required research engineer, research scientist or applied scientist profile rather than general software engineering or AI application integration.
A strong answer: Give your exact title, timeframe and research setting, then identify one or two modeling decisions, experiments or systems components for which you were directly responsible. Separate your contribution from the broader team’s work.
Where people lose it: Candidates lose points by describing a company’s AI product without identifying any research, experimentation or model-level work they personally performed.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

STEM AND CODING on Sidequest

What stem and coding AI work pays

Rates by specialty, the platforms hiring, and every open role in this field.