Research and learning

Post-training and preference optimisation

Post-training and preference optimisation is the work of taking a pretrained model and shaping how it behaves. Teams use demonstrations, reward signals and reinforcement learning to make a general model dependable at the specific things a product needs.

A pretrained model rarely behaves the way a product needs it to. More companies now want to shape that behaviour themselves rather than accept defaults. The hires are tricky because the skills sit between research and engineering. Strong candidates run clean, reproducible comparisons and can tell real gains from noise. Many come from fine-tuning, experiment design or evaluation, and the best are sceptical of their own results.

Hiring now 3

Roles in this family

3 roles

Hiring nowPost-Training Research EngineerA Post-Training Research Engineer adapts pretrained models into ones that behave as the company needs. They turn preference data and reward signals…Hiring nowReward Modelling ResearcherA reward modelling researcher builds models that learn what good output looks like from preference comparisons. They give post-training teams a…Hiring nowReinforcement Learning Research EngineerA Reinforcement Learning Research Engineer improves how a model behaves by training it against interactive signals. They matter because reward gains…

Where the people come from

Model fine-tuning, experiment design and evaluation

Not to be confused with. Large-scale pretraining belongs to Foundation model pretraining.

Core skills

  • Deep learning
  • Reinforcement learning
  • Preference learning
  • Model evaluation

Where it is used

  • Cross-sector AI platforms

More in Research and learning

Learning theory and statistical machine learning0 rolesFoundation model pretraining0 rolesReasoning, planning and search0 rolesLanguage and multilingual intelligence0 rolesComputer vision and spatial perception0 rolesSpeech, audio and music intelligence0 rolesMultimodal and generative world models0 rolesCausal inference, decisions and experimentation0 roles

← All new job families

Start a search

Hiring in post-training and preference optimisation?

Tell us the work you need done. We find the people already doing it, whatever their job title says.