Research Scientist — Epistemic Reasoning

About the role

Aletheia is our kernel-level truth verification subsystem. It's what separates Primordia from systems that confidently hallucinate — every reasoning chain passes through a verification layer that checks claims against stored knowledge, flags contradictions, and assigns confidence scores that propagate through the system.

This isn't prompt engineering. This is applied epistemology at the architecture level. You'll be advancing the fundamental research that underpins Primordia's position that AI systems should be verifiable, not just impressive.

What you'll do

• Research and develop formal verification methods applicable to neural reasoning chains
• Build probabilistic confidence scoring systems that propagate through multi-step reasoning
• Design contradiction detection algorithms that work across modalities and time
• Collaborate with the Mnemonic team to ground claims in persistent knowledge stores
• Publish research where appropriate and represent Primordia in the epistemic AI community
• Translate research advances into production-ready components

What we need

• PhD in AI/ML, Computer Science, formal logic, or epistemology — or equivalent research track record
• Deep familiarity with uncertainty quantification and Bayesian methods
• Experience with formal verification or automated theorem proving
• Strong publication record or equivalent demonstration of research output
• Ability to move between foundational research and applied implementation

Nice to have

• Background in philosophy of mind, epistemology, or formal semantics
• Experience with neuro-symbolic systems or hybrid AI architectures
• Prior work on AI interpretability or mechanistic interpretability
• Knowledge of AI safety literature and alignment research

The question of how AI systems can verify their own reasoning is one of the most important open problems in the field. If that's the problem you want to work on, reach out.