Research Engineer (Reinforcement Learning)
LiveKit · Remote · North America · APJ · EMEA ·
COMPENSATION
$135k – $300k
Listed on the employer's job board (Ashby)
- Fully remote role
- Python + Reinforcement Learning
- Post-training & Fine-tuning
- Work with open-weight models
- Senior team collaboration
LiveKit provides the infrastructure for voice-driven computing, powering applications for major tech companies. In this role, you will focus on post-training challenges such as long-horizon context retention and reliable tool use in live conversations. You will own the pipeline from synthetic data generation to model deployment, ensuring agents perform well across voice and text channels.
What you'll do
- Build training environments and verifiers for model development
- Manage synthetic data pipeline from generation to quality checks
- Conduct end-to-end training experiments and analyze results
- Develop evaluation criteria required for model releases
- Select and adapt open-weight base models for specific tasks
- Ensure trained behaviors work for both voice and text agents
- Deploy models to production and iterate based on real usage
What you'll need
- Strong proficiency in Python engineering
- Experience taking models from raw data to production
- Treat data as a product focusing on coverage and leakage
- Design rewards to prevent model exploitation of weaknesses
- Comfortable working with GPUs and understanding their limits
- Ability to decide when training is necessary versus unnecessary
- Effective collaboration in a remote work environment
Nice to have
- Experience with post-training, fine-tuning, or RL like GRPO
- Familiarity with TRL, verl, OpenRLHF, or custom training loops
- Knowledge of vLLM, SGLang, or multi-GPU training with FSDP
- Background in training tool-using or multi-turn agents
- Experience building execution sandboxes or eval harnesses
- Work with open-weight families like Qwen or Llama and LoRA
Stack
- Python
- TRL
- verl
- OpenRLHF
- vLLM
- SGLang
- FSDP
- Qwen
- Llama
- LoRA
How you'll work
- Fully remote position
- Collaborate with a small, senior team
What you get
- Competitive salary and equity package
- Health, dental, and vision benefits
- Flexible vacation policy
- Opportunity to shape a developer platform brand