Melbourne, Australia / Australian Permanent Resident
Senior Data Scientist & ML Engineer · PhD in Search & Recommendation Systems. I build production AI across the stack: search ranking, retrieval pipelines, experimentation and measurement, content understanding, RAG, and agentic automation — and I explain it to the people who have to use it.
My background starts in classical IR and ML — PhD-level research on personalised ranking and search, then four years as a Data Scientist at Redbubble building search, recommendation, and content-understanding systems over 100M+ user events. It runs through to the modern AI layer: LLM orchestration, RAG, agentic workflows, evaluation engineering, and CI/CD-grade deployment.
I'm strongest where retrieval quality, ML system design, and data engineering matter together: search that has to be measurable, recommendations that have to survive a real catalog, and agents that have to work reliably across messy enterprise data. The other half of the job is communication — I've owned experiment readouts for product stakeholders, taught 12+ lectures to a bootcamp cohort, and run onsite discovery with clinicians and contract reviewers who don't speak engineering.
The work I go deepest on: search ranking and retrieval, the experimentation foundation that proves a model moved something, and content understanding at catalogue scale. These are employer systems rather than public repos — each case study opens with its system map, experiment design, and measured impact.
Data Science · Search & Recommendation
Semantic retrieval and ranking for a marketplace with 100M+ user events
Marqo · GCP Vertex AI · BigQuery · Python · Vector Search · MLOps · Offline Evaluation · A/B Testing
Data Science · Experimentation
Making A/B results trustworthy — GA4 to BigQuery, offline eval, ship gates
A/B Testing · GA4 · BigQuery · Experiment Design · Metric Design · Offline Evaluation · Python · SQL
Data Science · Content AI
Language-image models for classification, moderation and duplicate detection
Language-Image Models · Image Embeddings · Classification · Anomaly Detection · Python · Taxonomy Design · SEO
Applied AI products, data-science case studies, agentic developer tooling, and client engagements — original builds, collaborations, and research forks kept clearly apart.
Open Projects Dashboard →Healthcare AI · Contract Intelligence
Document AI for Australian Healthcare Contracts
Python · FastAPI · Pydantic · Claude Sonnet · Azure Doc Intelligence · Tesseract OCR · Docker
Forward-Deployed · Client Engagement
Operations system for an occupational therapy practice — built onsite
Next.js · TypeScript · Supabase · PostgreSQL · Row-Level Security · Google Routes API · Geospatial Clustering
Voice AI · Collaboration
AI phone assistant — transcripts, summaries, risk analysis, hold-for-me
Next.js · TypeScript · WebSocket · Realtime Audio · LLM Summarisation · PostgreSQL
Data Science · Search & Recommendation
Semantic retrieval and ranking for a marketplace with 100M+ user events
Marqo · GCP Vertex AI · BigQuery · Python · Vector Search · MLOps · Offline Evaluation · A/B Testing
Agentic Developer Tooling
Remote orchestration + multi-agent control plane for AI coding agents
React · TypeScript · Express · Socket.IO · xterm.js · SQLite · PM2 · Git Worktrees · PTY
Teaching · Curriculum Design · LangGraph
6-week AI engineering course project — curriculum design + hands-on instruction
Python · FastAPI · LangGraph · Groq · Pydantic · Redis · MCP · NeMo Guardrails · RAGAS · Docker · Render
Most of what I build has to be handed to someone who didn't build it. That has meant teaching a 13-student cohort to ship a deployed multi-agent system, running 12+ hands-on lectures for engineers moving into AI, introducing agentic workflows to accounting professionals with no engineering background, and scoping two client systems from onsite interviews with clinicians and contract reviewers. Translating a technical system into the language of the people who own the problem is a skill I practise deliberately, not a side effect.
Teaching · Curriculum Design · LangGraph
6-week AI engineering course project — curriculum design + hands-on instruction
Audience — Career-change and early-career engineers with no prior agentic-AI experience
Teaching · Applied AI Engineering
Guest Instructor at Australia IT Group — RAG, Agents, Fine-tuning
Audience — Working engineers moving into AI engineering — mixed backgrounds, no shared ML baseline
Workshop · Non-technical Audience
Agentic AI for accounting professionals — a non-technical audience
Audience — Accounting professionals — domain experts with no engineering background
Discovery · Stakeholder Communication
How I scope a client system before writing any of it
Audience — Non-technical domain experts — clinicians, practice admin, contract reviewers
AI Engineer on CWBot and TriageAgent. Architected stateful multi-turn LLM agents with LangGraph and MCP, designed RAG workflows over enterprise knowledge bases, implemented guardrails and evaluation pipelines, instrumented request tracing and token/cost accounting, and owned CI/CD and deployment standards for production AI services.
Two forward-deployed engagements as sole engineer. Ran discovery with clinicians and admin staff before scoping an occupational therapy practice's operations system, including route-aware scheduling via geospatial clustering. Delivered document AI for Australian medical contract automation in a regulated healthcare setting, designed so output was checkable rather than trusted blindly.
Contribute runtime and serving optimizations for LLM inference backends used by agentic systems, spanning backend runtime, distributed serving, and prompt-language features for models including DeepSeek R1, Llama 3 and Qwen.
12+ bootcamp sessions on production RAG, vector search, multi-agent systems, QLoRA fine-tuning and RAGAS evaluation. Designed Dispatch.AI, a 6-week project taking 13 students to a deployed AI booking assistant. Delivered a hands-on agentic AI workshop for accounting professionals, adapting technical content for a non-technical business audience.
Led search and recommendation enhancements with Marqo vector search and GCP Vertex AI MLOps pipelines, improving add-to-cart by 0.5% and CTR by 10%. Owned production ML workflows end to end — offline evaluation, experiment design, deployment coordination and post-launch metric review — communicating results and model logic to product stakeholders. Designed ML infrastructure over 100M+ user events and drove the GA4 to BigQuery analytics migration underpinning A/B testing.
Shipped production content-classification and moderation systems using language-image models, reducing IP moderation risk and improving operational throughput. Built image duplicate-detection pipelines and data-quality/anomaly-detection workflows, and ran tagging, taxonomy and SEO experiments to improve product discovery.
Web Search and Data Mining Group. Developed Autosub, a collaborative subtitle-generation system using speech-to-text APIs, and contributed to web page mining projects using data mining and time-series analysis.
PhD in Computer Science — Search & Recommendation Systems, RMIT University. B.Sc. in Computer Science, Harbin Institute of Technology.
VLDB Journal, DASFAA, ADC. Research focus: personalized ranking, preference adjustment, and continuous summarization.
ADC Best Student Paper Award, DASFAA Best Student Paper Runner Up, Google Code Jam Top 1000, ACM-ICPC Asia Regional Silver Medal.
Google Cloud Professional Data Engineer, Microsoft Azure AI Fundamentals, GCP data lake / warehouse modernization, deep learning and NLP specializations.
My strongest fit is where search, retrieval, and ML engineering meet production AI: meaningful evaluation metrics, large or noisy catalogs, data pipelines that need to be reliable, and systems that need to improve measurably over time — not just demo well once. PhD-trained in search and recommendation, hands-on in data science and MLOps, now building LLM-powered systems end-to-end — and able to explain any of it to the people who have to live with it.
Start a conversation