We’re creating a position to push DeepSeek-R1-style breakthroughs into advanced search and retrieval. Our goal: build an autonomous agent that reasons across knowledge bases—leveraging the file ingestion, vector search, graph construction, and sophisticated retrieval flows in our open-source R2R platform. What You’ll Do Prototype a reasoning-driven system combining R1 & Claude/Gemini/4o to navigate and interpret millions of documents. Distill successful methods onto smaller models for efficient deployment. Scale Up to power our next-gen system after demonstrating real performance gains. Ideal candidates should have Ph.D.-level expertise or equivalent experience in AI, machine learning, or related fields, with a passion for reasoning, retrieval, and experimentation. This is a unique opportunity to shape the heart of R2R into a solution that truly “thinks” about the data it processes.