Research Engineer

Role Summary

You build Aidan himself. You empower Aidan to orchestrate his abilities across four modalities in realtime: voice (two-way, multilingual conversation), hands (live browser use inside real products), eyes (proactive vision on the user's screen), the Brain (a self-improving context graph), and the verifiers (how we can keep evaluating Aidan's performance in real scenarios). Making that feel human is one of the hardest engineering problems in AI, and our engineering team's unique ability to solve it is what makes Sable special.

Responsibilities

• Improve agent runtime across topics like voice latency, perception, personalization, action planning for live browser use
• Build out the Skills system: turning screen recordings and human-led demos into reliable, durable knowledge of how to operate a customer's UI
• Advance perception: improve Aidan's ability to see and understand the user's screen in realtime and use those visuals to decide when and how to intervene
• Strengthen the improvement engine: study Aidan's interactions to determine where he took a suboptimal action, what he could have done better, and how he can use those moments to learn and improve automatically

Requirements

• Engineer with deep experience in at least one of: realtime systems (voice/streaming/media), LLM agents and orchestration, browser/computer use, and applied ML such as multimodal evals
• Comfortable in a small team where you own problems end to end and communicate fluidly
• Preferred: previous experience at a leading AI company or institution

About the Company

ABOUT SABLE

Sable built Aidan, the first AI employee who can lead customer calls using realtime voice, vision, and browser use. Aidan runs a live, two-way conversation inside a real product environment, clicking through the product like a human, watching the user's screen, and adapting the journey on the fly. Every conversation feeds a self-improving context graph we call the Brain, so Aidan gets smarter with each call.