SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sable has built Aidan, an AI employee capable of leading customer calls using realtime voice, vision, and browser interaction. Aidan operates as a live, two-way conversational agent within real product environments, clicking through interfaces like a human, observing user screens, and dynamically adapting interaction flows. Each conversation feeds into a self-improving context graph called the Brain, enabling Aidan to improve continuously.
In this role, you will translate performance criteria into clear, defensible evaluation signals and use those signals to drive improvements to Aidan's capabilities. You will design and execute evaluations across a broad frontier of areas including voice interaction, vision processing, and browser automation. Your work will directly inform optimization of the agent, its context system, and underlying models.
Key responsibilities:
- Design evaluations for voice, vision, and browser use that direct and accelerate improvements in the agent, context, and models
- Build and maintain internal benchmarks and simulation systems
- Develop systems to score customer-specific scenarios and identify capability gaps
You will work closely with the team to ensure evaluation frameworks are rigorous, scalable, and aligned with real-world performance needs.
Requirements:
- Engineering background with demonstrated statistics and/or applied machine learning experience
- Comfort working in production codebases
- Bonus: prior experience designing or working with evaluations, or experience with voice, realtime, or browser/computer use agents
