Gerardo EzequielSpatial intelligence for cities
CV
← All work

Leadership / Product · Completed engagement

Related portfoliosIndustryFounder / CTO

CommunityAI

Decision boundary · schematicThree inputs meeting one bounded mandate. The engagement is confidential, so this plate shows a role, not a place.

Technical leadership for a community-intelligence venture that helps teams work with public neighbourhood data.

What existsAn open-data proof of concept, a technical roadmap and spatial validation controls for the founding team.

My role
Fractional Chief Technology Officer
Period
2026, ended October
Status
Completed engagement
Focus
Organisation and platform

The question

The founding team needed to turn varied public neighbourhood datasets into a dependable product. Sources had to remain traceable, models needed meaningful geographic tests, and area-level signals had to stay distinct from customer outcomes.

My contribution

As fractional CTO, I began with a technical audit and turned its findings into a staged implementation roadmap. I set technical strategy, designed the solution and data architecture, and led data engineering, bespoke spatial modelling and software delivery. The open-data proof of concept explored audience geography, location choices and media allocation. I introduced source and licence records, spatial holdouts, baseline comparisons and reproducible release checks, keeping customer-specific validation separate.

How I approached it

  1. Review the foundations

    I audited the existing technical work and set out a staged platform roadmap.

  2. Rebuild the data layer

    I connected public sources to documented processing, licence records and model inputs.

  3. Test before release

    I introduced spatial holdouts, baseline comparisons and checks separating research outputs from customer-specific claims.

What exists

  • A technical audit and staged implementation roadmap informing the fractional CTO engagement.
  • An open-data architecture with documented sources, processing and model provenance.
  • Spatial holdouts and baseline comparisons built into validation.
  • Release controls distinguishing research outputs from customer-specific validation.

Where the work stands

The engagement ended in October 2026. This case describes my delivered technical work, not ownership or proven customer impact. Detailed architecture and results remain confidential. Predictive signals do not establish causal effects.

Design choices and constraints
  • Use spatial holdouts and agreed baselines to test model performance.
  • Serve aggregate area-level evidence, outside personal-data workflows.
  • Require customer outcome data before claiming added value for that customer.
  • Keep confidential architecture, counterparties and commercial findings private.

An early-stage product, a collaborative team and confidentiality obligations limit what can be shared publicly.