Scaling Science Researcher

Also called Scaling Laws Researcher

A Scaling Science Researcher works out how model performance changes with compute, data and model size. Their estimates tell a company how to spend a training budget and what results to expect before committing to a large run.

What they do

They design small and mid-sized training experiments that can be compared fairly. They fit curves to the results, test whether the fits hold at larger scale, and write down where they stop being reliable. They work with architecture, data and infrastructure teams to decide how to split compute between model size and training data. They check that runs are reproducible, that data quality and provenance do not distort the trends, and that noise is measured rather than ignored. Much of the week goes on planning sweeps, reading results sceptically and rerunning what looks odd. The output is a study others can trust and act on.

What a strong hire delivers

In the first months, a strong hire delivers a scaling study for the current model recipe, with clear uncertainty ranges and stated limits on extrapolation. They set up a repeatable experiment set that others can rerun. They give a concrete recommendation on compute allocation, and they check earlier predictions against what the larger runs actually produced. The team gains a shared way to judge whether a change helps at scale.

How you can tell they are good

Ask them to walk through a scaling study they ran. Look for how they chose the experiment grid, how they measured noise, and what they predicted before the larger run finished. Ask where their estimates were wrong and why. A strong candidate talks about out-of-sample error, confounds in data and training setup, and the limits of their own extrapolations. Be wary of tidy curves with no error bars.

Why it is hard to fill

Few people have run scaling experiments end to end, because few organisations have had the compute to try. The work often sits inside titles like research scientist, pretraining researcher or research engineer, so job titles say little. Candidates tend to be either strong researchers with thin experimental discipline, or strong engineers with little statistical rigour. The mix is rare, and the best people are usually already close to a large training budget.

Where to find them

Look in pretraining and foundation model teams at frontier labs and well-funded model builders, where scaling work is often done by research scientists or research engineers. Academic groups in deep learning, optimisation and statistics also produce candidates, especially those who have published reproducible training studies. Some come from infrastructure or evaluation teams who have moved closer to experiment design. Titles vary widely, so read the work, not the label.

Indicative pay

Not enough published evidence yet to give a fair range for this role, which is common for work this new. We benchmark it for you as part of a search.

Interview questions for a Scaling Science Researcher

Describe a scaling study you ran. How did you choose the model sizes, data amounts and compute points?
A strong answer explains the grid design, the reasons for each range, and what budget limited. It covers repeated runs and noise. Worry if they only describe fitting a curve to whatever runs were available.
How did you test whether your scaling estimates were right?
Look for held-out larger runs, predictions recorded in advance, and honest reporting of error. They should name cases where the estimate failed. Worry if validation was only in-sample fit quality.

4 more questions, with what a strong answer covers, are in the Role Kit.

Job description you can post

We are building foundation models and need someone to tell us, with evidence, how performance changes as we scale compute, data and model size. You will design the experiments, fit and test the trends, and shape how we spend our training budget.

What you will do

Design scaling experiments across model size, data and compute
Fit scaling relationships and quantify their uncertainty
Validate predictions against larger training runs
Recommend how to allocate compute between model size and data
Work with architecture, data and infrastructure teams to keep runs comparable
Document studies so others can reproduce and challenge them

What you will bring

Demonstrated experience training deep-learning models with reproducible results
Strong grounding in experimental design and statistical reasoning
Hands-on work with transformer models in PyTorch or JAX
Understanding of data quality, deduplication and provenance effects
Evidence of predictions checked against outcomes
Clear written communication of results and their limits

Nice to have

Experience with large distributed training runs
Published or shared work on scaling or training dynamics
Experience with evaluation design for large models

Free to use and adapt. If it saves you time, a link back to this page is appreciated.

Skills that matter

Deep learningAdvancedBuild and diagnose trainable neural models
Transformer modellingAdvancedImplement attention-based sequence or multimodal models
Experimental designWorkingChoose comparisons that identify credible effects
Data quality and provenanceWorkingTrack dataset origin, validity and transformations

Other roles in foundation model pretraining

Hiring nowPretraining Research EngineerA pretraining research engineer runs controlled experiments on how large models learn from broad data. The work decides which architectures, data…Hiring nowModel Architecture ResearcherA Model Architecture Researcher tests changes to how a model is built, to improve its behaviour or efficiency. The work tells a company which…EmergingTraining Dynamics ResearcherA Training Dynamics Researcher works out why large training runs wobble, stall or diverge, and fixes the cause. Fewer failed runs means less wasted…

Related roles in other families

Hiring nowMachine Learning Research ScientistLearning theory and statistical machine learningA Machine Learning Research Scientist develops and tests learning methods against explicit hypotheses. They give the company evidence about what…Hiring nowStatistical Learning ScientistLearning theory and statistical machine learningA Statistical Learning Scientist builds prediction methods that show their assumptions and their uncertainty. For a company shipping models into…Hiring nowLearning Theory ResearcherLearning theory and statistical machine learningA Learning Theory Researcher works out what learning algorithms can and cannot do, and proves it. For an AI company, that means firmer ground for…Hiring nowAutomated Machine Learning ResearcherLearning theory and statistical machine learningAn Automated Machine Learning Researcher builds search procedures that choose and tune models within a fixed compute budget. It matters because…

How we wrote this

Role definitions are our own editorial work, checked against occupational classifications, standards, research and employer pages. Titles vary by company, and this is a description of the work, not a job advert.

Careers: jobs, Anthropic
Transformers documentation, Hugging Face

← Foundation model pretraining All new job families

Start a search

Hiring a Scaling Science Researcher?

Tell us the work. We find the people already doing it, whatever their job title says, and you meet them already interviewed for your role.