Open-source deep research MCP. Qwen3-30B-A3B-Thinking via OpenRouter, cited web synthesis for Claude Code, Codex, Cursor, Hermes and any MCP-compatible agent.

SAGE: Self-Adaptive Goal-directed Executor — A multi-tool LLM agent with DAG-based hierarchical planning, ReAct reasoning, and evidence-guided self-correction for automated research synthesis. Built from scratch, no frameworks.

Agent skills for systematic review, meta-analysis, umbrella review, and AI-assisted evidence synthesis.

A Python toolkit for systematic literature reviews — by Proportione. PRISMA 2020 + MMAT 2018 + bibliometrics, with CLI and Streamlit demo.

(experimental) Parallel multi-agent swarm architecture for collective AI intelligence. All agents live. All agents share memory. All agents iterate until consensus.

Demonstration of the Framework to manage Variance Theories in Software Engineering

Enable parallel multi-agent consensus research with a live shared memory and preserved dissent using a three-tier hierarchical model.

Naturalistic fMRI corpus (100 top-tier papers, 2021-2026) + NotebookLM-integrated research synthesis pipeline. 20 expert queries across 7 categories surface SOTA, outstanding questions, and AI applications.

Turn messy context into a brief you can think with for Claude and Codex.

A curated map of judgment infrastructure for human-led AI.

Consolidate multiple AI deep research reports into one verified master document. Gated systematic review pipeline, zero runtime AI dependency.

Turn research corpora into auditable, evidence-graded claim inventories. A Claude skill with archaeology and medical-scientific grading presets.

Aligning opposites into unified intelligence.

Cut LLM costs by 10–30% using better orchestration—not model changes. A visual demo comparing Weft-style pipelines vs map-reduce and full-buffer baselines, showing token usage, cost savings, and % reduction.

Replication package for the paper "Theory Building from Data Strategy Studies: Aggregating Evidence on Model Quantization in Deep Learning Systems".

Versioned, machine-readable research synthesis you operate, not paste — builds an evidence ledger, enforces traceability invariants, and never asserts an insight the data doesn't support.

PhD Lecture "Research Synthesis: An Introduction to Reviews, Meta-Analyses, and PRISMA Criteria"