Praveen Suthaharan
AI Safety Researcher · Cognitive Neuroscientist
Human-centered artificial intelligence

I build safe AI systems to support the human mind.

CV

Portfolio

Education Active

Metacognitive learning

Interpretable end-to-end multi-agent feedback workflow.

Step 1 Human reflections text input Step 2 MP-constrained LLM engine autonomous scorer Step 3 Score & feedback output for learner Learner longitudinal reflections Day 1 · Day 2 Day 3 · Day 4 Day 5 · Day 6 scoring_agent.py Analyze reasoning Sufficient info? Flag elaboration MP Judges framework Comprehension · Judgement Evaluation · Final decision Confidence probabilistic scores Evidence binder spans · rubric links Self-checks consistency · alignment Score & feedback generator detective_agent.py EDF Step 1 · Detect weakest dimension low score uncertainty no trend diagnostic_agent.py EDF Step 2 · Diagnose cause of weakness weakest_dimension: Evaluation primary_tag: weak_evidence_linking scaffold_agent.py EDF Step 3 · Scaffold selection reasoning tag scaffold library prompt style dimension personalization_agent.py Personalization layer profile · prior knowledge · UX EDF Step 4 Personalized advisory targeted learner support Yes · Not now · No
Mental health Coming soon

Future system