Research Intern

Company Pramaana Labs is a frontier AI lab building the verification layer for AI. Founded in 2025 and headquartered in Palo Alto, we turn complex human knowledge, including tax codes, legal rules, clinical guidelines and security protocols, into a formal representation, so every AI answer can be traced, challenged, and proved. Pramaana's architecture pairs foundation models trained to formalize and reason with a symbolic world model encoded in Lean 4. Pramaana was founded by a team out of Google, DeepMind, and Glean, combining frontier AI researchers and formal methods experts. The Role As a Research Intern at Pramaana, you’ll contribute to one of several areas: model training, harness engineering, AI for Formal research or benchmark development. You’ll help run training experiments, investigate new approaches to symbolic reasoning, or build evaluation harnesses that help us measure and improve model performance. What You'll Do Help run model training experiments and analyze the results. Designing harnesses for auto-formalization and prover systems. Build and maintain benchmarks to measure performance. Investigate where models succeed or fail, and suggest improvements. Work with researchers and engineers to turn findings into the next set of experiments. You Will Thrive in This Role If You Have A solid grasp of deep learning fundamentals and how to evaluate models. The ability to run careful experiments, communicate findings clearly, and work well with a research team. Technical depth in at least one area relevant to the role, such as model training, formal methods, or evaluation systems. The initiative to make progress on open-ended problems with limited direction. Close attention to detail in your code, data, and results. Especially Strong Candidates May Also Have A background in formal verification (Lean, Coq, or Isabelle) Hands-on experience training or fine-tuning LLMs, including work on pretraining, supervised fine-tuning, or reinforcement learning. A strong publication record in top-tier venues (NeurIPS, ICML, ICLR). #J-18808-Ljbffr