Murciel Igor

At Africa Quantitative Sciences, I’m the sole architect and engineer behind DataGoat, an LLM-backed analytics platform that lets non-technical users query datasets in natural language using DuckDB SQL or sandboxed Python and R.

I designed DataGoat’s memory-efficient Parquet-on-object-storage serving architecture, delivered a 319,875-row preview in 860ms, and built streaming Excel ingestion roughly 12x faster than openpyxl. I also built privacy controls that redact PII before prompts leave the local server and restore it in streamed responses.

I built DataHub, Rwanda’s public health open-data portal commissioned by the Ministry of Health and coordinated by the WHO Rwanda office. I extended CKAN with a custom Python extension, a two-level taxonomy, secure embedded Superset dashboards, and a config-driven white-label platform model.

I also developed secure data-collection capabilities for AfriCollect, including JWT authentication, department-level permissions, AES-256-GCM PII encryption, and MongoDB reporting. My work spans PostgreSQL, Python, FastAPI, Docker, AWS, data pipelines, and production systems designed to stay efficient on constrained infrastructure.