LLM Council
LLM Council is a synthetic extension of research conducted in the Network Effects Lab and the Spatial Networks Lab at Arizona State University's W.P. Carey School of Business, working with Stanford professors Dr. Timothy J. Richards and Dr. Yueming (Lucy) Qiu.
Our work established that structured peer consensus, or disagreement, produces stronger judgment and influence than any single source. We developed experimental methodology for measuring how judgment propagates through trust networks and how behaviours diffuse through spatial networks. Later Dr. Mary L. Shelman of Harvard invited me to present my research.
The core finding was consistent: peer networks outperform anonymous single-source judgment by a factor of three.
That was 2013.
A decade later, AI evaluation faces the same problem. Single-model judges inject systematic bias, what we now call hallucination. The solution is the same.
LLM Council is the direct evolution of this research: a synthetic network designed to outperform any single frontier model.
AI Architect with 16 years building AI and data capabilities across regulated organisations. Research background in predictive modelling and discrete choice models (Arizona State), deep learning and NLP (Indian School of Business), and LLM fine-tuning (QUT). 73+ citations on Google Scholar.
Experience
Qchat (35,000 government users, 7,000 daily active) and Corella (student/teacher AI assistant rolling out statewide 2026). Azure OpenAI, RAG, vector search, constitutional AI guardrails.
Microsoft 365 Copilot rollout for 3,000 users. Azure AI Foundry and AWS Bedrock platforms. Enterprise AI policy and guardrail frameworks.
RAG-based prioritisation engine on Databricks using vector embeddings. Enterprise analytics architecture on Azure Synapse and Fabric.
Multi-country platform for Pacific Labour Mobility. Azure Cognitive Search and document management.
Research: "Fine-Tuning GPT-3 and BART for Context-Aware Summarisation of Social Media Conversations"
AI Applications and Policy Research Lab. Deep learning, spatial econometrics, and social network analysis research.
Machine learning and spatial analytics solutions.
Enterprise architecture and analytics for card services.
Predictive analytics, discrete choice demand models, and retail scanner data modelling at W.P. Carey School of Business.
Publications & Talks
73+ citations on Google Scholar
Research areas: Transformer Architecture, NLU/NLG Fine-tuning, Multi-agent Consensus Systems, Social Network Analysis
- Transformer fine-tuning strategies for natural language understanding and generation (QUT)
- GPT and BART model optimisation for context-aware text generation
- Social network effects on peer consensus and decision-making
- Multi-stage deliberation frameworks for improved AI judgment
- Dec 2025 - CSIRO Responsible AI Symposium, Adelaide: "Grounding Global Standards in Australian AI Governance"
Certifications
- Azure AI-102 (AI Engineer), AI-900, DP-900
- ISO 42001 Lead Implementer
- TensorFlow, NLP, CNNs - Google Brain / DeepLearning.ai (2019)
- Certified Scrum Master (2013)
Brisbane, Australia