🐧 Unify-Agent: An end-to-end unified multimodal agent for faithful, knowledge-grounded image generation.

Structured ecosystem of 190+ AI systems spanning foundation models, agentic reasoning, reinforcement learning, generative architectures, scientific ML, and self-evolving neural systems. A research-driven lab exploring scalable intelligence.

Reference implementation for "StateTrace: Explicit State Modeling for Verifiable Multimodal Rule Reasoning." State-bottlenecked reasoning framework with typed state schema, constrained action decoding, transition verification, and dense transition-level rewards for multimodal sports rule adjudication.

Reference implementation for "RefTrace: Grounded Normative Reasoning via Retrieval-Augmented Agentic Graph-of-Thought for Multimodal Video Adjudication." Agentic retrieval-loop architecture with Game-State Temporal Hypergraph (GSTH), 6-action typed grammar, Normalized Trace Reward (NTR), and two-stage SFT→GRPO training for NFL penalty adjudication.

Adaptive parallel reasoning skills for Claude Code.

System for monitoring and evaluating the reasoning processes of large language models. It streams model outputs, extracts intermediate reasoning steps, and uses a secondary model to detect logical, computational, and process-level errors. The goal is to improve reliability by analyzing how answers are derived, not just the final output.

Pipeline to investigate structured reasoning and instruction adherence in multimodal LLMs

Cross-task diagnosis of Capacity Domination in ReAct agents — paper artefacts and reproduction code.

Causal reasoning engine for KETER-PRIME-KDIF the "why" layer of Agentic Symbiotic Systemic Intelligence.

An architectural blueprint for cognitive mind filtering and semantic synthesis, structuring human intent into deterministic knowledge networks.