🛠 Tech Stack

💼 About This Role

You'll design the long-term memory architecture for LLMs, working with a world-class team towards self-improving superintelligence. You'll advance the field through open publishing of research and open-source code. This role requires being in-person 5 days a week in downtown San Francisco.

🎯 What You'll Do

• Define key abstractions for the LLM memory layer
• Build memory architectures supporting multiple memory types
• Research memory sharing between multiple agents
• Improve context management to solve long-context derailment

📋 Requirements

• Deep expertise in LLMs and retrieval
• Track record of impactful research (publications or open source)
• Ability to balance execution speed with empirical rigor

✨ Nice to Have

• Experience with multi-agent systems
• Published memory-related AI research

🎁 Benefits & Perks

• 🚀 Work on frontier AI at a small, high-impact team
• 📝 Open publishing culture with papers and open-source code
• 🏢 In-person collaboration in downtown San Francisco

📨 Hiring Process

Initial screen (30 min), technical screen (1-1.5 hours), paid in-person work trial (2 days onsite in SF).

🚩 Heads Up

• Requires 5 days/week in-office with no hybrid option
• Explicitly states 9-5 work style guarantees failure
• High ambiguity and pressure in a small team environment