Startups · AI

Member of Technical Staff - Mechanistic Interpretability

Vmax · San Francisco · On-site

← All jobs
About Vmax

Vmax is automating reinforcement learning. We transform proprietary data and evals into new sets of environments. We refine agents on new examples of the tasks they are intended to perform. Backed by South Park Commons.

About the role

LLMs are fantastically powerful and there is a rapidly growing corpus of work devoted to understanding their internal representations and computations. We use the tools of mechanistic interpretability to enhance reinforcement learning by generating intrinsic rewards as a supplement or alternative to downstream human-generated verifiers.

What they're looking for

  • PhD or equivalent experience in machine learning, reinforcement learning, or a closely related field
  • Track record of research excellence, as demonstrated by publications, open source work, deployed AI systems, or other substantial technical contributions
  • Deep understanding of modern machine learning, especially reinforcement learning, representation learning, and large language models
  • Strong familiarity with LLM post-training methods
  • Experience designing and running rigorous ML experiments, including ablations, baselines, evaluation design, and failure analysis
  • Expertise with Python and at least one major ML framework such as PyTorch or JAX
More about this role

V max is an applied research lab developing AI capable of open-ended learning. We are building systems to exceed humans in all capacities by optimising beyond the local maxima of learning from human expertise.

LLMs are fantastically powerful and there is a rapidly growing corpus of work devoted to understanding their internal representations and computations. We use the tools of mechanistic interpretability to enhance reinforcement learning by generating intrinsic rewards as a supplement or alternative to downstream human-generated verifiers.

  • Develop methods for using mechanistic interpretability to extract useful training signals from the internal states of language models.
  • Turn representations, features, circuits, and causal model behaviors into intrinsic rewards for reinforcement learning.
  • Compare interpretability-derived rewards against human feedback, learned reward models, verifiers, and task-level outcome rewards.
  • Design metrics and baselines for reward quality, including alignment with intended behavior, generalization across tasks, robustness, and resistance to reward hacking.
  • Investigate how internal representations evolve during RL and post-training, and use...

Read the full posting on Vmax's site ↗

Research and Engineering

Build your edge while you search

Free tools for founders and investors, plus VC Unfiltered, our take on startups, venture and the people who build them.