Gabriela Olivera

Data analyst — Mount Maunganui, NZ

Hi, I'm Gabriela.

I spent fourteen years in operations and procurement before moving into data, which means I know where operational numbers break before they ever reach a report.

I build the whole thing end to end: pipelines, models, and dashboards people actually use.

See the work
Gabriela Olivera

4 government APIs→ spatial join across 22,834 parcels→ weighted scoring model→ 6-page dashboard

Terroir

Screenshot of the Terroir Streamlit dashboard's suitability map, showing kiwifruit land-suitability parcels colour-coded across the Bay of Plenty coastline

Where should kiwifruit grow in Bay of Plenty?

Geospatial land-suitability analysis

  • 22,834
    Parcels analysed
  • 21,491
    Scored
    of 22,834 parcels
  • 13,040
    Candidates
    of 16,414 Excellent

I pulled four live sources — LINZ cadastral parcels, S-map soils, LCDB land cover and Open-Meteo climate — into a Python pipeline, joined them spatially, and scored every rural parcel in four districts for kiwifruit suitability.

Case study Live dashboard Code

The live dashboard runs on a free host and can take about 30 seconds to wake up on first visit — it isn't broken, just warming up. The case study loads instantly.

Apophenia

Screenshot of the Apophenia simulator interface, showing operational-risk sliders and forecast charts for a kiwifruit export season

What happens when an export season goes wrong?

Operational risk intelligence simulator

  • 9
    Live variables
  • 26wk
    Forecast
    with a fixed-width projection band
  • 4
    Data feeds
    each independently guarded

I built a simulator in hand-written JavaScript, Chart.js and Mapbox GL JS where nine operational variables — weather, port congestion, fruit quality, regulatory load — recalculate risk, cost of delay and a 26-week forecast live, against four independent public APIs.

Live simulator Code
“On Terroir I found a bug that had misclassified 19.4% of parcels. I caught it because a filtered subset came out larger than the group it was filtered from — which is impossible. Every fix like that is in the public commit history.”
— Gabriela, on the Terroir case study

How I work

Python SQL JavaScript Power BI Streamlit geopandas Git