Selected work

Case studies built around evidence and delivery decisions.

This section separates verified professional work from synthetic portfolio studies. Each case states its evidence kind and limitations.

Verified professional work

Analytical pipeline modernisation

Migration of a recurring analytical process from a legacy SQL Server implementation to a modular Databricks and PySpark workflow with automated quality controls.

PySpark
Databricks
SQL Server
SQL
Data quality
Statistical validation

Read case study

Synthetic portfolio study

RAG Reliability study

A proprietary product direction for deciding when a RAG system should answer or defer, supported here by a synthetic answer/defer study.

Python
uv
pytest
BM25
sentence-transformers
all-MiniLM-L6-v2
reciprocal-rank fusion
SHA-256 artifact verification

Read case study

What each case study documents

Business and delivery context.
Technical approach and deliverables.
Evidence and result classification.
Constraints and limitations.
Production and handover considerations.