Project

Singapore PDPA Legal Compliance RAG Agent

A retrieval-augmented legal workflow that turns cross-document compliance comparison into a reusable question-answering system.

Python · RAG · Prompt Engineering · OpenAI API · Legal NLP

2026-present

Product Snapshot

Role: Product R&D and algorithm engineering
Focus: RAG architecture for legal consulting, retrieval design, prompt chaining, and PDF parsing research.
Validation: Partner law-firm project records report a 33% improvement in the system benchmark baseline.
Public Proof: RAG workflow, evaluation materials, and partner benchmark records; the metric was not independently rerun for this portfolio.

Next Step

Portfolio PDF About the Founder Email

Context

In 2026, I worked at Lanyue Intelligence on a RAG agent for Singapore PDPA legal-compliance consulting.

The core problem was practical: lawyers had to compare multiple regulatory documents, trace the relevant clauses, and turn them into an explainable recommendation. The manual process was slow, expensive, and inconsistent.

What I owned

I helped shape the system beyond simply attaching a model to documents:

supported the RAG architecture for legal consultation
helped design parsing and retrieval flows so clauses could be recalled more reliably
adjusted multi-step prompt chains to preserve legal context while reducing improvisation
aligned system behavior with real consulting needs instead of demo-friendly outputs

Result

Partner law-firm project records report a 33% improvement in the benchmark baseline. The metric was not independently rerun for this portfolio.

That result is useful evidence that the system could reduce friction and improve consistency in a high-precision workflow, while the source qualifier remains visible.