Research and learning

Reasoning, planning and search

Reasoning, planning and search is the research that makes models better at multi-step problems. It covers how a system works through a task, plans actions within limits, and spends extra effort at inference time. AI companies need it because capable models still fail on long chains of logic.

Teams have found that a bigger model alone does not fix unreliable multi-step work. Attention has moved to how a model thinks through a problem, and how much effort it spends on each one. Hiring is tricky because the field is young and titles vary. Candidates often come from algorithms, ML research or operations research, so check they can show results on hard problems, not just fluency in the theory.

Hiring now 1Emerging 3

Roles in this family

4 roles

Where the people come from

Algorithms, ML research or operations research

Not to be confused with. Agent integration focuses on external actions rather than core reasoning methods.

Core skills

  • Transformer modelling
  • Reinforcement learning
  • Model evaluation
  • Optimisation and operations research

Where it is used

  • Cross-sector AI platforms
  • Physics and mathematics

More in Research and learning

← All new job families

Start a search

Hiring in reasoning, planning and search?

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