We are building systems that reason over time, not just respond in a single turn. As a Cognitive Systems Research Engineer, you will explore how autonomous AI systems can plan, reflect, adapt, and improve across extended tasks.

This is applied research. You’ll investigate architectures inspired by cognitive science, distributed systems, and reinforcement learning — then translate them into practical implementations inside our agent framework.

Your job is not to publish papers. It’s to close the gap between theoretical reasoning capability and real-world system reliability.

What You’ll Do

• Design and prototype advanced agent architectures, including hierarchical planning, memory compression, reflection loops, and tool selection strategies.

• Experiment with multi-agent systems that coordinate toward shared goals.

• Develop structured reasoning frameworks that improve long-horizon task completion.

• Investigate memory strategies, retrieval augmentation, context window optimization, and state persistence.

• Build evaluation systems to measure reasoning quality, adaptability, and failure modes.

• Run controlled experiments comparing architectural approaches and quantify their tradeoffs.

• Collaborate with product and infrastructure teams to transition successful prototypes into production-ready systems.

• Continuously track emerging research and rapidly test promising ideas.

What We’re Looking For

• Strong engineering skills in Python or similar languages.

• Experience working with large language models and prompt orchestration.

• Understanding of planning algorithms, search strategies, reinforcement learning concepts, or distributed coordination.

• Comfort designing experiments and interpreting noisy results.

• Ability to move between abstract theory and concrete system implementation.

• Intellectual curiosity paired with pragmatism — ideas must eventually ship.

Nice to Have

• Experience with multi-agent simulations or emergent behavior systems.

• Background in ML systems, cognitive science, AI research, or complex adaptive systems.

• Experience building evaluation harnesses for probabilistic models.

• Familiarity with vector databases and retrieval-augmented generation systems.

Why This Role Matters

Most AI systems today operate reactively. True autonomy requires structured reasoning, memory, adaptation, and failure recovery across time.

You will help define how those capabilities are engineered — not imagined.

Location & Work Style

Remote. Full-time. High ownership. Experiment-driven.

Compensation

Competitive salary and equity based on experience and impact.