Shuyin Zhang

Public research record

Founder & CEO, NSP AI LABS INC. · AI Systems Researcher & Protocol Architect

Founder and independent AI systems researcher developing the Narrative State Protocol (NSP), an architecture for preserving explicit cognitive state, evidence provenance, and verification conditions across long-running AI workflows. Designed its domain-agnostic core, schema-defined domain layers, and deployment runtimes; authored three public preprints reporting system design, controlled evaluations, and observed limitations.

Research & company leadership

Founder & CEO — NSP AI LABS INC.

July 2026–Present

Canada · AI research and infrastructure company

  • Founded NSP AI LABS INC. to develop and steward NSP; lead research direction, protocol and product architecture, benchmark strategy, and commercialization planning.
  • Maintain NSP Core as retained proprietary IP while defining pathways for open benchmarks, evaluation protocols, and technically separated reference tools.

Creator & Lead Architect — Narrative State Protocol (NSP)

January 2026–Present

Independent research initiated January 13, 2026

  • Designed a three-layer protocol: domain-agnostic primitives for entities, perspectives, time, evidence, and constraints; schema-defined domain vocabularies; and deployment-specific runtimes.
  • Built evidence-linked state, declarative constraints, diagnostics, lifecycle gating, and persistent memory/history so assumptions, conflicts, blockers, and verification state remain inspectable across long workflows.
  • Created a structured long-horizon research runtime with persistent problem state, bounded exploration campaigns, critic review, and machine-readable evidence artifacts.

Bounded result: A self-authored controlled ablation across four mathematical tasks reported 85.0% full-system success versus 62.5% ablated (n=80; CMH p=0.021). No individual task remained significant after multiplicity correction; the result has not been independently validated.

Founder & Technical Director — Beijing Cloudtime Culture Consulting Co. Ltd.

2014–Present

Beijing (2014–2022) · Tokyo (2022–2024) · Canada (2024–Present)

  • Designed and delivered AI-enabled enterprise systems for Japanese businesses, including document OCR and archival, internal knowledge platforms, and customer-service agents.
  • Led localization engineering and software delivery for Tencent, Perfect World, NetEase, and Epic Games; built custom asset-extraction pipelines and performed authorized reverse engineering for licensed titles.

Publications

  1. Structure Over Similarity: Structural Cognition Across Mathematics, Software, and Music. Zhang, S.Y. · 2026 · Public preprint doi:10.5281/zenodo.20062417
  2. Programmable Emergence via Cognitive Engine: Longitudinal Evidence of AI Research Maturation. Zhang, S.Y. · 2026 · Public preprint doi:10.5281/zenodo.19392279
  3. Think Before You Act: A Cognitive Gating Protocol for Reliable AI Agents. Zhang, S.Y. · 2026 · Public preprint doi:10.5281/zenodo.19334052

Capabilities

Research Long-horizon AI systems, cognitive-state protocols, benchmark design, controlled ablation, evidence and constraint modeling
Architecture Schema-driven systems, knowledge graphs, lifecycle middleware, persistent state, critic and verification workflows
Engineering Python, JavaScript, HTML/CSS, SQL, YAML/JSON schemas, FastAPI, Flask, React, Gradio, SQLite, pytest
Languages Chinese (native), English (fluent)

Education

B.A. Fine Arts — LuXun Academy of Fine Arts, Shenyang, China 2009

Public application CV · Direct contact details are supplied privately to program administrators.