Machine Learning Scientist I/ii, Scientific Reasoning

Lila Sciences Lila Sciences · AI Frontier · One Charles Park, Cambridge, MA · Physical Sciences AI

Machine Learning Scientist focused on Scientific Reasoning, designing novel frameworks for LLM-based reasoning, exploring in-context learning and self-reflection, building scalable model prototypes, and integrating domain knowledge into reasoning systems.

What you'd actually do

  1. Design and formalize frameworks for scientific reasoning with LLMs, including structured prompting, reasoning chains, and test-time compute.
  2. Explore and implement methods for in-context learning, self-reflection, and adaptive reasoning in scientific discovery workflows.
  3. Build scalable model prototypes that can be deployed to solve frontier scientific problems.
  4. Collaborate with scientists and engineers to encode domain knowledge into reasoning systems that integrate symbolic and statistical approaches.

Skills

Required

  • Python
  • LLM frameworks (PyTorch, HuggingFace Transformers, LangChain, LlamaIndex)
  • LLM reasoning methods (in-context learning, test-time compute, chain-of-thought, or tool-augmented reasoning)
  • theoretical research
  • practical ML engineering

Nice to have

  • causal reasoning
  • symbolic AI
  • probabilistic programming
  • open-source LLM reasoning frameworks
  • scientific discovery pipelines
  • multimodal reasoning
  • publications in top ML/AI conferences

What the JD emphasized

  • scientific reasoning with LLMs
  • in-context learning
  • self-reflection
  • adaptive reasoning
  • scalable model prototypes
  • domain knowledge
  • symbolic and statistical approaches
  • LLM reasoning methods
  • in-context learning
  • test-time compute
  • chain-of-thought
  • tool-augmented reasoning
  • theoretical research
  • practical ML engineering
  • scalable solutions

Other signals

  • design novel frameworks for scientific reasoning with LLMs
  • implement scalable frameworks that integrate with Lila’s platforms
  • encode domain knowledge into reasoning systems that integrate symbolic and statistical approaches