San Francisco startup building open-ended task generation for autonomous systems.
Member of Technical Staff - RL Algorithms
You'll spend your days designing and implementing reinforcement learning algorithms that power systems which generate their own tasks, rather than relying on static human-defined objectives. The core challenge is coaxing models to explore skill spaces with genuine autonomy—crafting reward structures, curiosity mechanisms, and credit assignment schemes that scale beyond narrow domains. Expect to run experiments on clusters, dig into why an agent fixates on trivial loops, and rewrite your hypothesis before lunch. This is a small team where your paper-reading habit and your PyTorch debugging sessions both matter.
What they're looking for
- 5+ years developing RL algorithms in production research environments, with published work or shipped systems to show for it
- Deep expertise in PyTorch and distributed training infrastructure for large-scale experimentation
- Track record with exploration, self-supervised skill discovery, or open-ended learning paradigms
- Fluency translating research insights into clean Python, not just prototype notebooks
- Comfort with ambiguity: the problem definition shifts as the systems get smarter