Research landscape / 07 active projects
Systems at the edge of intelligence.
Live system
OutputGuard
LLM output safety API. Validates model outputs before they reach users.
API Live
outputguard.dev →
01 / LLM Security
OutputGuard
LLM output safety API. Validates model outputs before they reach users.
Open research →
02 / LLM Security
A-PRIME
Mechanistic interpretability. Causal localization of tool-call drift in Llama-3.1-8B.
Open research →
03 / LLM Security
AIAT
Adversarial red-teaming framework for LLM evaluation.
Open research →
10 / Intelligence
Intelligence
Open signals turned into evidence, maps and research direction.
Explore field →
11 / Publications
Publications
Preprints on tool-call interpretability and agent harness security.
Read publications →Public observatory / skill-radar
Field Intelligence
robotics ML
HOT1,000pub velocity / 30d
robotics ML is moving quickly, with 1,000 publications tracked in the latest 30-day window.
large language models
HOT1,000pub velocity / 30d
large language models is moving quickly, with 1,000 publications tracked in the latest 30-day window.
AI governance
HOT915pub velocity / 30d
AI governance is moving quickly, with 915 publications tracked in the latest 30-day window.
AI safety
HOT693pub velocity / 30d
AI safety is moving quickly, with 693 publications tracked in the latest 30-day window.
AI policy
HOT663pub velocity / 30d
AI policy is moving quickly, with 663 publications tracked in the latest 30-day window.
mechanistic interpretability
HOT632pub velocity / 30d
mechanistic interpretability is moving quickly, with 632 publications tracked in the latest 30-day window.
AI alignment
HOT588pub velocity / 30d
AI alignment is moving quickly, with 588 publications tracked in the latest 30-day window.
quantum_ml
OPEN1,097pub velocity / 30d
quantum_ml is gaining momentum as new work broadens the field across the latest 30 days.
ai_auditing
OPEN853pub velocity / 30d
ai_auditing is gaining momentum as new work broadens the field across the latest 30 days.
cybersecurity_ai
OPEN754pub velocity / 30d
cybersecurity_ai is gaining momentum as new work broadens the field across the latest 30 days.
formal_verification
OPEN752pub velocity / 30d
formal_verification is gaining momentum as new work broadens the field across the latest 30 days.
federated learning
OPEN279pub velocity / 30d
federated learning is gaining momentum as new work broadens the field across the latest 30 days.