Research Engineer (Reinforcement Learning)

LiveKit is hiring a Research Engineer to build post-training for its voice and text agents.

The interesting problems here are the long-horizon ones: staying useful across many sessions, working with context that accumulates over time, and using tools reliably in the middle of a live conversation. You would build the environments and verifiers the models train against, and own the synthetic data pipeline from generation onward. It is reinforcement learning applied to agents that are already in production rather than to a benchmark.

LiveKit powers voice AI for OpenAI, xAI, Salesforce, Coursera and Spotify, collectively facilitating billions of calls each year.