I'm Ferhat Erata, an Applied Scientist at AWS AI, working on Agentic AI and Neurosymbolic AI for Code, Security & Reasoning.
- π PhD in Computer Science from Yale University (Advisor: Prof. Ruzica Piskac)
- π My current research focuses on:
- Learning Randomized Reductions (ICML 2026 Spotlight: https://arxiv.org/abs/2412.18134)
- Behavioral Alignment of Specification and Code
- Semantic Program Compression (Library Learning)
- Training transformers to solve NP-complete problems with RL
- Neurosymbolic methods for AI safety β combining neural and symbolic reasoning to prevent hallucinations
- Test-time compute strategies and regime-aware uncertainty estimation for LLM reasoning
- π Currently exploring the foundations of learning theory, program synthesis, and formal verification
- π¬ Ask me about automated reasoning, program analysis, SAT/SMT solving, and neuro-symbolic AI
- π« How to reach me: https://ferhat.ai

