Fractional CAIO in Pomona, CA

Enterprise data architecture at LERETA — and what it has to do with AI

This page is built on a real engagement: I served as Fractional CTO for LERETA, the Pomona-based real estate tax services company, for approximately four years — leading a $20M enterprise architecture and data modernization across a 30+ person engineering team.

To be clear: this was architectural and modernization work, not an AI engagement. I did not lead AI initiatives at LERETA.

What I did build is four years of deep, hands-on experience with the data systems, pipeline architecture, and data governance patterns that AI capabilities depend on — and that perspective is what I bring to AI strategy advisory for data-intensive companies in this market. The most common reason enterprise AI projects stall is not the model — it’s the data: siloed, inconsistent, inaccessible, and not designed for programmatic use. The LERETA work addressed all of that. Understanding those problems from the inside — having spent four years solving them — is what makes AI readiness assessments and strategic advisory for this type of company substantive rather than generic.

What AI-ready data architecture actually looks like

Enterprise architecture modernization and AI readiness aren’t two separate projects. They’re the same project, described differently.

An AI model needs to be trained on data that is clean, consistent, and representative. It needs to run inference against data that is accessible in real time or near-real time. It needs to operate inside a system that can capture feedback — model predictions, user corrections, outcome signals — to support continuous improvement. Getting all of that in place requires exactly the work that a serious data architecture modernization produces:

• Consolidated data sources — removing the silos that make it impossible to train a model on the full picture
• Normalized data models — consistent schemas that a model can actually learn from, rather than fighting through data quality issues at inference time
• Modern pipeline architecture — replacing brittle ETL processes with pipelines that support both batch training and real-time serving
• Documented data lineage — knowing where data comes from and what transformations it has gone through, which is a governance requirement for any production AI system

The LERETA modernization addressed all of these. The company emerged from that engagement with data infrastructure that a machine learning team can actually use.

The AI opportunity in real estate data

Real estate and property tax data is one of the richest AI application environments in the enterprise software space:

Automated valuation models (AVMs). ML models trained on historical sales prices, tax assessed values, property characteristics, and neighborhood signals can generate property value estimates at scale and speed that manual approaches can’t match. For a company sitting on deep real estate data, AVMs are the highest-value, most immediately buildable AI use case.

Document extraction at scale. Real estate transactions generate enormous volumes of documents: deeds, liens, easements, tax records, title commitments, legal descriptions. LLMs fine-tuned or RAG-architected on this domain can extract structured data from unstructured documents at a fraction of the cost and time of manual review.

Tax data anomaly detection. Tax assessment data contains patterns — value changes, ownership transfers, exemption activity — that signal fraud, data quality issues, or market dislocations. ML models trained to detect these patterns add a continuous audit layer that rule-based systems can’t match.

Natural-language query interfaces. Non-technical staff querying complex real estate and tax databases currently rely on pre-built reports or SQL-fluent colleagues. An LLM layer that translates natural-language questions into structured queries opens that data to the whole organization.

These are near-term, high-ROI applications sitting on top of the data architecture the LERETA modernization built.

The Inland Empire AI landscape

Pomona anchors the western edge of the Inland Empire — one of California’s fastest-growing business regions, with an AI adoption curve just beginning to accelerate:

• Logistics, supply chain, and distribution — the IE is the logistics backbone of Southern California, with massive distribution infrastructure that is a natural fit for AI in route optimization, demand forecasting, inventory management, and autonomous warehouse operations.
• Real estate and property services — a large base of title, tax, appraisal, and property management companies serving the region’s enormous residential and commercial real estate market.
• Healthcare and clinical services — a growing healthcare system serving a large population, with AI applications in diagnostics, documentation, and operational efficiency.
• Manufacturing and industrial — the IE’s significant manufacturing base is beginning to look at AI for quality control, predictive maintenance, and process optimization.

The common thread is data-intensive operations — businesses that run on structured information and complex workflows where AI can create measurable efficiency.

What a Fractional CAIO delivers for an Inland Empire firm

The highest-value deliverables for most Pomona / Inland Empire companies:

1. AI readiness assessment — data audit, process inventory, infrastructure gap analysis, and governance readiness evaluation. The output is a precise picture of where you are and what it takes to get to production AI.
2. Data architecture for AI — modernizing data infrastructure specifically for AI workloads: pipeline architecture, data model normalization, feature stores, and real-time serving capabilities.
3. AI use-case roadmap — a prioritized map of where AI creates the most value in your specific business, with build/buy/API recommendations and an ROI model for each.
4. LLM strategy for data-intensive industries — document extraction, natural-language query, and domain-specific model fine-tuning for real estate, logistics, and industrial applications.
5. ML model design and oversight — AVM design, demand forecasting, anomaly detection, and other predictive models for data-rich environments.
6. AI governance framework — data quality standards, model monitoring, audit requirements, and the governance structures that make AI adoption responsible and durable.

These are detailed on the main Fractional CAIO services page — substantiated here by four years of data architecture leadership at a Pomona real estate data company that produced exactly the AI-ready infrastructure most companies are still trying to build.

How the engagement works

• Discovery (2–4 weeks): AI readiness assessment — data audit, infrastructure gap analysis, process inventory, and use-case prioritization. Output: a written AI roadmap and readiness report.
• Foundation phase (if needed): data architecture upgrades specifically scoped to AI readiness — pipeline modernization, schema normalization, feature store design.
• AI build phase: use-case architecture, model design, LLM integration, and automation workflow design for the priority initiatives.
• Ongoing: model monitoring, data quality management, and roadmap updates as AI capabilities and business requirements evolve.

If you’re a Pomona or Inland Empire company evaluating AI strategy — whether you have a solid data foundation or are still building one — the next step is a discovery call.