Projects
Industrial production systems I've delivered, each with a short write-up on the problem, how it was built, and what changed as a result.
This is a selected set of projects, not an exhaustive list.
GenAI / Agents / Agentic Flows
Strategic Environment Analyser Agent
Monthly manual sweep → continuous agent over 250+ sources
A four-stage agent that scans for regulatory change, tests each lead against a financial organisation's compliance baseline, and routes confirmed changes through human approval.
- Agentic workflows
- Multi-stage pipeline design
- Human-in-the-loop
- Airia.ai
- Claude Haiku 5.5
- +4
Agentic RFP & Vendor Evaluation
Replaced a ~3-month manual tender review with an agentic workflow; ~4 hours of reviewer time saved per vendor response
A three-stage agentic workflow for procurement: completeness checking, rubric-based scoring and side-by-side vendor comparison, with human approval gates at every decision point.
- Airia.ai
- LLM-as-a-judge
- Human-in-the-loop
- Guardrails
- Prompt engineering
- +1
Large-Scale Unstructured Notes Processing
1M+ free-text notes and 1B+ tokens turned into a structured data space during a platform migration
A cloud LLM pipeline that summarised and extracted implicit and explicit fields from over a million unstructured member-conversation notes, built to feed a structured data model during a company-wide platform transition.
- AWS Bedrock
- Step Functions
- Lambda
- S3
- Claude 3.5 Sonnet
- +2
Intelligent Legal Document Processing
AUD 220k saved against vendor quotes: 10,000+ contracts and 130k+ pages processed for about AUD 5k of infrastructure
An automated extraction pipeline pulling 20+ structured fields from more than 10,000 heterogeneous contract documents, built to support a legal team contract cleanup that had been quoted as an outsourced project.
- AWS (S3, Textract, Lambda, Step Functions)
- Azure OpenAI GPT-4o
- LLM-as-a-judge
- Fuzzy scoring
- Apttus & Salesforce REST APIs
- +1
Multi-Agent System for Incident Resolution
Agents consolidate context across structured and unstructured sources for 500+ incidents a day
A multi-agent system that connects to live and historical incident data, then compares, summarises and consolidates it, cutting the manual overhead of gathering context before an incident can be resolved.
- Microsoft TaskWeaver
- Vanna.ai
- RAG
- Text-to-SQL
- Vector databases
- +3
Organisation-Wide RAG Chatbot
RAG assistant rolled out to 800+ employees, with SSO auth, GraphRAG retrieval and persistent session memory for day-to-day work with sensitive internal documents
An internal, SSO-integrated chatbot over SharePoint and Confluence content, using GraphRAG with reranking and DynamoDB-backed session memory so staff can securely query and act on sensitive documents.
- RAG
- GraphRAG
- Reranking
- AWS Bedrock
- DynamoDB
- +2
Statements QA Automation
350+ automated checks replaced brittle Excel macro QA
A document intelligence pipeline that extracts key fields from year-end financial statements and cross-validates them against member database records.
- LLM-as-a-judge
- Document intelligence
- Deterministic rule sets
- Fuzzy scoring
- AWS
- +1
Cooee: AI Chatbot for Energy Analytics
Natural-language access to a billion-point, 200+ table live energy data space, for both internal engineers and the public, without writing a query
A chatbot for a university energy analytics platform that grew from a single NLU/generative pipeline into a multi-agent RAG architecture serving both an internal, database-connected assistant and a public one over open documentation.
- LLM agents
- Text-to-SQL
- RAG
- Contextualized QA
- Vector databases
- +6
Data Science
Emotion & Theme Analysis of Call Recordings
60,000+ support calls transcribed and analysed into dashboards on emotion, theme and outcome
An AI framework for a health support call centre that transcribes call audio, then detects and summarises caller emotions, emotion transitions and themes, correlated against metadata to show where services could improve.
- Azure Cognitive Services
- Topic modelling
- Emotion detection
- Text classification
- Keyword extraction
- +3
Emotion Analysis of Financial Earnings Call Transcripts
12,521 earnings-call transcripts (2010–2020) analysed for CEO and analyst emotion, linked to stock returns
A finance-domain emotion analysis pipeline over more than 12,500 earnings-call transcripts, tracking emotional agreement and conflict between speakers and correlating the results against market variables.
- Python
- NLP
- Word embeddings
- Deep emotion extraction (Emotion AWARE)
- Document parsing
- +2
Emotion AWARE: A Framework for Comprehending Emotion in Text
A neuro-symbolic framework producing multi-granular, intensity-scored emotion profiles, robust to negation and modifiers, explainable by design
An AI framework that elicits adaptable, robust and explainable emotion profiles from text, combining a fine-tuned language model with a rule-based lexicon and embedding-space matching, and underpinning the call-recording and financial-transcript emotion work.
- Python
- BERT
- Deep learning
- Text/sentence embeddings
- Lexicon construction
- +2
Lead Generation Pipeline
Automated organisation-profile construction from a dozen-plus public web sources, each field attributed and confidence-scored
A web-scraping and information-extraction pipeline that builds structured organisation profiles (address, contacts, leadership, industry) by crawling and cross-referencing public business directories, search results and company websites.
- Python
- Flask
- Web scraping
- MongoDB
- Azure (VMs, Storage Queues)
- +1
Patient Segmentation for Targeted Health Campaigns
Patient segments and their significant variables identified from combined appointment and outcomes data, driving AI-generated campaign emails per segment
A patient segmentation pipeline that combined appointment records and Patient-Reported Outcome Measures to find behavioural patterns, then used the resulting segments to drive targeted, LLM-generated email campaigns.
- Association rule mining
- Clustering
- Classification
- Azure ML Studio
- OpenAI
- +2
Energy Nowcasting & Forecasting for LEAP
Nowcasting and forecasting models built on 15-minute smart meter data to predict energy usage, solar generation and emissions for a live energy analytics platform
Nowcasting and forecasting models over 15-minute-frequency smart meter data from the La Trobe Energy Analytics Platform (LEAP), predicting near-term energy usage, solar generation and emissions ahead of consumption.
- XGBoost
- LightGBM
- Neural networks
- LSTM
- Time series forecasting
- +1
Human Activity Classification from Accelerometer Data
CNN models classifying and predicting activity levels over time from large-scale accelerometer data, before and after knee surgery
A CNN-based pipeline, built for a University of Auckland research collaboration, classifying and predicting activity levels from large-scale accelerometer datasets of pre- and post-osteoarthritis knee surgery patients.
- CNNs
- Time series classification
- Python
- Signal processing