AI Developer · Auckland, NZ

Production AI, measured.

I build AI agents and automations that run in a real business every day, and I test them until I can show they work. Below are the numbers, and a demo you can try.

MRR 0.79

Best retrieval score across 36 reviewed test questions

Hit@5 97%

Questions with the right page in the top five results

Judge agreement 94%

How often my LLM reviewer agreed with my own review

Setups compared 12

Three retrievers, each at four chunk sizes

From my RAG Research Assistant. See the full results

Live demo

Ask it something.

Code on GitHub

RAG Research Assistant

Ask questions about five well-known papers on retrieval and RAG. It answers only from the papers and cites the paper and page for every claim.

  • Says so when the papers don't cover a question, instead of guessing.
  • Hybrid search: keyword and meaning-based retrieval combined.
  • Tested against a reviewed question set, with tests running on every push.

Python · LangChain · BM25 · sentence-transformers · OpenAI · Gradio · pytest · GitHub Actions

Live on Hugging Facerag-research-assistant

Try asking

What happens when the answer sits in the middle of a long context?

The demo runs on Hugging Face and may take a few seconds to wake up.

Open full screen Answers cite paper and page

Analytics work

Data turned into decisions.

All projects

What I build

From the data to the agent.

Agents and RAG

Agents that classify, retrieve and reply, and question-answering over documents.

ClaudeOpenAILangChainTool callingHybrid retrievalOllama

Evaluation

Checks that catch bad output before it reaches a customer.

LLM-as-judgeDeepEvalRAGASLangSmithpytest

Automation and integration

Workflows joining e-commerce, inventory, shipping, helpdesk and social platforms.

n8nJavaScriptRESTWebhooksDockerLinux

Data and analytics

Dashboards and analysis that turn raw data into decisions.

SQLPythonRPower BIDAXPostgreSQLSASMATLAB

Experience

How the toolkit grew.

Full CV
Skills used in each role, oldest to newest
Skill area
Databases and warehousing
Dashboards and reporting
Statistics and machine learning
Optimisation and AI search
Automation and integration
LLMs and agents
Evaluation and testing
Servers and deployment

Click a column to see the role.

Master of Analytics (Honours)Auckland University of Technology, 2023 to 2024. Thesis: classifying stress from EEG data, 86.8% cross-validated accuracy.
Bachelor of Technology, Computer Science EngineeringJawaharlal Nehru Technological University, India, 2007 to 2011
SAS Certified SpecialistBase Programming Using SAS 9.4

Contact

Let's talk about what you're building.

Open to AI and automation roles in Auckland.