Published portfolio view — clean read
Résumé·Herndon, VA

Résumé

I build data and orchestration platforms that outlive me. Four platforms across three industries since 2015: housing finance, telecom, commercial data products. Two went to production, and my own firm tried to buy the first one from the client. One beat Palantir Foundry head-to-head. Two are in build now. I’m the hands-on architect on every one of them.

01

Experience

May 2026
Head of Platform Engineering · Accenture (Associate Director) — greenfield data and AI platform for a commercial data-products business
May 2026 — present · Herndon, VA (remote)
  • Lead platform engineering for a greenfield data and AI platform on Databricks and AWS: forty people in six squads.
  • Responsible for platform selection, target architecture, and the interface contracts between squads.
  • Lead the engineering teams building the platform's first data products.
Apr 2026
Founder and architect · Shipper — AI software-delivery loop (TrueDataLab)
Apr 2026 — present · Independent
  • One agent drafts the requirement with me, a second builds it, and an isolated third tries to break it against a fixed rubric. Bounded retries, then merge and deploy.
  • 1,730 tasks across 17 repositories since April 2026. The validator rejected 18.8% on substance, and 27.7% of model spend goes to verification.
  • I set the requirements and make the calls that are expensive to reverse. Claude writes the code.
Sep 2019
Platform Architect · Accenture (Senior Manager, then Associate Director) — governed self-serve data engineering platform, Tier-1 US telecom
Sep 2019 — May 2026 · Herndon, VA
  • Designed and led the build (2022 onward) with a 4–7 engineer team: ~4,500 production DAGs processing ~75B records/day for 14 orgs and 300 peak users, on Airflow and BigQuery.
  • Designed the user contract — analysts write SQL or config, never code — and enforced it at authoring time rather than in review: one DAG, one table; 28 reusable DAG types; sub-60s click-to-production.
  • API-first years before MCP existed, so an AI agent became a first-class operator via 81 MCP tools without redesign.
  • Won a head-to-head evaluation against Palantir Foundry. Still in production.
  • Led the Oracle-to-BigQuery migration on the platform: ~2,000 tables, ~2,500 pipelines and 150 dashboards in seven months, as routine changes through the standard promotion workflow.
  • Predecessor scheduler (2021, solo): metadata-driven orchestration in Oracle with dependency resolution in SQL, four runner types, file-wait, retries, per-job validation, and a parallel bulk data mover with per-stage instrumentation.
2023
Creator · Natural-language analytics co-pilot — telecom crowdsourced data
2023 · Solo, six weeks of evenings
  • Question to SQL to narration to chart on GPT-4, built weeks after function calling shipped and before JSON mode existed: routing via function calls, a SQL error-feedback retry loop, chart specifications via model-written code.
  • Productized by my employer under its own brand.
2014
Sole engine author; engagement lead · Accenture (Associate Manager to Manager) — test data platform and DataOps, federal housing-finance GSE
2014 — 2019 · Washington, DC
  • Wrote a metadata-to-SQL compiler in PL/SQL: one generated INSERT-SELECT per table with referential integrity, correlated lookups, per-step audit, versioned metadata, archive and restore. 1–2 billion rows per run; over a trillion cumulative across three years.
  • The engine outlived two user interfaces — an Access front-end, then a Spring Boot product built by a separate team on the contract I owned. In 2017 my own firm tried to buy it from the client.
  • Lead architect for Delphix data virtualization across 200+ applications; XML data generation enabling integration between two federal organizations.
  • Led an 11-person test data management engagement and built a DataOps program (virtualization, test automation, generation, validation, CI/CD) supporting a 10 TB warehouse.
2012
Systems Analyst, then Associate Manager · Accenture
2012 — 2014 · São Paulo, Brazil
  • Designed a data integration framework (Oracle, Datastage, Shell, Java) for an international bank; coordinated a team of six.
  • SAP migration: coordinated 140+ data extraction and load processes across Finance, HR, Fleet, Compliance and BI with a team of 11 at peak; trained a Datastage class of nine.
2010
Senior BI Analyst · Sky Brasil
2010 — 2012 · São Paulo, Brazil
  • Owned the sales department's critical SSIS data loads and the primary BI server hardware migration; built proactive failure detection for load configuration errors.
2002
Programmer to Analyst Programmer · Accenture and Interativa Multimídia
2002 — 2010 · São Paulo, Brazil
  • PL/SQL data warehousing from the start: departmental DW for a US pharmaceutical company, partitioned and sliding-window load tuning at an international bank, Datastage-to-Informatica reimplementation, and a four-developer PL/SQL team lead.
  • Before that, Oracle and SQL Server installation, relational modelling, PL/SQL, and Delphi and ASP applications as an intern turned programmer.
02

Skills

Data platform architectureOrchestration design (metadata-driven; Airflow)Data modeling & schema governanceLakehouse (Databricks, Unity Catalog, Iceberg)SQL at scale (BigQuery, Postgres)PL/SQL (Oracle, 20+ yrs)PythonAgentic systems (MCP, agent orchestration and verification)GCP · AWSFastAPIContainers & CI/CDObservability (Prometheus, Grafana, Loki)Self-hosted infrastructurePortuguese (native) · English (fluent)
03

Education

Universidade Presbiteriana Mackenzie, São Paulo — B.S. Information Systems (Bacharel em Sistemas de Informação) · 2003 — 2006

04

Certifications

CBIP — Certified Business Intelligence Professional — TDWI · 2014