← All work

HUB24 · Data Enablement Lead

Private AI benchmarking platform

A privately hosted LLM application giving executives explainable natural-language analytics across 600,000 SMSF funds.

600,000
SMSF funds covered
~$400bn
AUM benchmarked

Overview

Executives wanted to ask benchmarking questions in plain English, with answers they could trust. I architected and delivered a privately hosted LLM application that combines:

  • a custom retrieval layer over fund data,
  • deterministic business logic for the numbers that must be exact,
  • domain heuristics encoding how SMSF benchmarking actually works, and
  • guardrails that keep answers grounded and minimise hallucinations.

The result is explainable natural-language analytics across 600,000 SMSF funds (~$400bn AUM).