RAG Proxy

Local AI that knows our policies, and keeps our data inside the building

TechKnowFile 2026 · University of Toronto · May 6-7

87.5%

RAG Accuracy

50.4%

Fine-Tuning

Monthly Cost

100%

Local & Private

Browse all 20 slides from the TKF 2026 presentation as a thumbnail grid.

How six failed fine-tuning runs led to a 730-line Perl solution that actually works.

What was tried, what went wrong, and what each failure taught us.

The full technical architecture: vector search, proxy interception, and the code.

AI and RAG terms explained in plain English. No jargon, no hand-waving.

Research papers, tools, and references from the talk.

The full timeline from first fine-tuning attempt to production deployment.

About Michael Limberger, the person behind the system.

Michael Limberger · Senior Web & Linux Administrator · Faculty of Dentistry