Ready to tackle the number one problem facing widespread AI adoption?
Join a pioneering company developing the world's first Agent to supervise Agents, LLMs and AI systems. As we move toward an increasingly autonomous AI future, scalable oversight isn't just important—it's essential for safe AGI deployment.
You'll work on the world's first Agent which supervises Agents, LLMs and AI systems. Gone are the days of basic evaluations—this completely automates the evaluation process and automatically debugs AI agents, giving intelligent feedback on how to fix any failures, planning issues or systemic issues. While AI engineers currently spend hours combing through agent traces searching for planning mistakes and tool use errors, this technology makes the process instantaneous and comprehensive.
This role offers an opportunity to solve fundamental research problems at the intersection of LLM post-training, reinforcement learning, and agents. You'll work on post-training, creating realistic RL environments for agent evaluation, building the next generation of agentic evaluation systems and developing proprietary benchmarks for emerging AI capabilities.
Your research focus:
- Core Research impacting their main Agent platform
- Develop cutting-edge benchmarks that evaluate LLM and agent capabilities as they emerge
- Create challenging RL environments that organically evaluate agents under realistic conditions
Advance agentic evaluation methodologies beyond traditional LLM-as-a-Judge approaches - Build and improve evaluator models that provide systematic oversight of autonomous systems
- Publish research findings in collaboration with top AI research labs
The role balances research with implementation, allowing you to see your theoretical contributions deployed in production systems used by industry leaders. You'll work directly with the founding team to shape the technical direction of AI supervision technology. They also co-publish papers with top research labs.
You should have:
- Strong hands-on research experience in LLM post-training or RL, or agent research
- Publication record at top conferences (NeurIPS, ICML, EMNLP, ACL, ICLR)
- Experience with LLM fine-tuning and understanding of transformer architectures
- Track record of following recent research and implementing state-of-the-art methods
- PhD in Computer Science or related field (preferred but not required)
- Ability to move quickly and thrive in a startup environment
You'll join a small but exceptionally high-calibre team working on problems that will define the future of responsible AI systems and AI safety. The company has proven market demand with existing enterprise customers (OpenAI, NVIDIA, Meta and more) and strong backing, allowing you to focus on breakthrough research with real-world impact.
Your package includes:
- Salary up to $300,000 (negotiable based on experience)
- Significant stock in a fast-growing company
- Comprehensive benefits, 401k, and unlimited PTO
- Relocation assistance available and sponsorship
You'll work from San Francisco, NYC also considered.
Ready to help ensure AI systems are safe and reliable as we approach AGI?
If you're excited about solving the most critical challenge in AI safety while publishing cutting-edge research, we'd love to hear from you. All applicants will receive a response.
Location: | San Francisco Bay Area, |
---|---|
Job type: | Permanent |
Emp type: | Full-time |
Salary type: | Annual |
Salary: | negotiable |
Job published: | 22/08/2025 |
Job ID: | 33119 |