EmergingContinual, adaptive and federated learning

Meta-Learning Researcher

Also called Learning-to-Learn Researcher

A Meta-Learning Researcher studies how models carry learning from one task to the next, so they adapt quickly to new tasks with little evidence. That matters when data is scarce, shifting or held by others.

What they do

They design methods that learn across a distribution of tasks, then test them on tasks the model has never seen. Day to day, that means building task sets, running sequential and federated experiments, and checking for drift as conditions change. They compare adaptation after a handful of examples against simple baselines. They also look at how uncertain the model is after adapting, and where privacy or distributed ownership limits what data can be shared. Much of the work is careful evaluation design, then writing up what held and what did not. The work is often done now under titles such as ML researcher or privacy engineer.

What a strong hire delivers

A strong hire delivers a working meta-learning method and an evaluation on unseen tasks that the team trusts. They set out how tasks are sampled and what counts as adaptation quality after limited new evidence. They show where the method beats simple baselines and where it fails. They leave behind reusable benchmarks, drift tests and a clear recommendation on whether the approach belongs in the product.

How you can tell they are good

Ask them to explain how they built a task distribution and why it matched the tasks the model would meet later. Ask what leakage between training and unseen tasks they guarded against. A strong candidate shows negative results with ease, compares against plain fine-tuning, and talks about uncertainty after adaptation. Be wary of anyone who reports only average gains on familiar tasks.

Why it is hard to fill

Hardly anyone holds this title. The work sits inside research, optimisation and privacy roles, so CVs rarely say meta-learning. Candidates often have depth in one strand, such as distributed optimisation, and less in another, such as evaluation on unseen tasks. Academic researchers may lack production experience. Engineers may lack the research rigour. We look for the mix, and we read projects rather than job titles.

Where to find them

Look at ML research groups working on few-shot learning, continual learning or transfer. Distributed optimisation and federated learning teams are another source, as are privacy engineers who have worked with decentralised data. They sit in university labs, frontier labs, applied research teams and companies that handle sensitive data across many owners. Check authors of papers on sequential benchmarks and drift.

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 Meta-Learning Researcher

  1. Describe a task distribution you designed for meta-learning. How did you decide it represented the tasks the model would meet later?

    Strong answers cover how tasks were sampled, how diversity was checked and how unseen tasks were kept separate. They admit mismatch risk. Worry if the answer is just a named benchmark with no reasoning about fit.

  2. How do you measure adaptation quality after only a few new examples?

    Look for fixed budgets of evidence, variance across tasks, and comparison with fine-tuning and training from scratch. Worry about single averages, tuned test sets or no baselines.

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

Job description you can post

We are looking for a Meta-Learning Researcher to help our models adapt quickly to new tasks with limited evidence. You will design methods, test them on tasks the model has not seen, and tell us honestly what works. Your findings will shape what we build into our product.

What you will do

  • Design and train meta-learning methods across task distributions
  • Build evaluations on unseen tasks with clear baselines
  • Test for drift in sequential and federated settings
  • Measure uncertainty after adaptation
  • Work with engineers to move useful methods into production
  • Document results, including negative ones, for the wider team

What you will bring

  • Demonstrated research or applied work in meta-learning, transfer or continual learning
  • Experience building task sets and evaluation protocols that avoid leakage
  • Working knowledge of distributed optimisation or federated learning
  • Ability to compare against simple baselines and report results honestly
  • Strong coding skills in a modern ML stack
  • Clear written explanations of methods and limits

Nice to have

  • Privacy engineering experience
  • Experience with uncertainty quantification
  • Published or open work on sequential benchmarks

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

Skills that matter

  • Continual learningAdvancedAdapt models while measuring retention and drift
  • Privacy engineeringAdvancedLimit disclosure and unauthorised personal-data use
  • Distributed systemsWorkingDesign reliable work across networked machines
  • Uncertainty quantificationWorkingEstimate confidence, calibration and model uncertainty

Other roles in continual, adaptive and federated learning

EmergingContinual Learning ResearcherA Continual Learning Researcher builds ways to update models on new data without losing what they already do well. It matters because retraining…Hiring nowFederated Learning EngineerA Federated Learning Engineer trains models across several data holders without pooling the raw data. That lets a company use data it cannot…Hiring nowOnline Learning ScientistAn Online Learning Scientist builds models that keep learning from arriving data under explicit feedback rules. It matters because products that…

Related roles in other families

Hiring nowComputer Vision EngineerComputer vision and spatial perceptionA Computer Vision Engineer builds detection and recognition systems that work in a defined setting, and proves how well they work. Companies rely on…Hiring now3D Vision ResearcherComputer vision and spatial perceptionA 3D Vision Researcher recovers spatial structure from images and depth data, then proves it holds up. Companies building robotics, mapping, AR or…Hiring nowVideo Understanding ScientistComputer vision and spatial perceptionA Video Understanding Scientist builds models that recognise actions, events and the dependencies between them in video. Companies hire for this…EmergingOpen-World Perception ResearcherComputer vision and spatial perceptionAn Open-World Perception Researcher builds vision models that know when they are looking at something unfamiliar. That matters because a system that…

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.

← Continual, adaptive and federated learningAll new job families

Start a search

Hiring a Meta-Learning 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.