HandMade{data}·Marcelo Thiesen
Private background note

Marcelo Thiesen

Founder HMData Since 2022
short bio

Private systems, internal applications, automation, and applied AI.

Marcelo Thiesen is a software and data engineer who builds and operates private data systems, internal applications, process automation, and applied AI workflows. An economist by training at Insper, with coursework in financial derivatives at the London School of Economics and an MBA in Business Analytics and Big Data from FGV, he began his career in financial modeling and M&A for infrastructure operators before moving into data engineering, analytics leadership, and custom software systems.

Since 2022, he has run HMData (HandMade {data}), an independent practice serving a small portfolio of companies as an embedded data and engineering partner. The work is typically ongoing rather than one-off, combining cloud infrastructure, business intelligence, process automation, custom applications, and applied AI built around each client's own operations.

His projects span healthcare, clinical research, logistics, property management, media analytics, professional services, and sports performance. Many engagements involve confidential, personal, financial, or operational data, handled in dedicated, access-controlled environments with privacy controls appropriate to the work.

background

From financial modeling to data engineering and custom systems.

Economics, analytics, and financial derivatives

BSc in Economics, Insper (2014). MBA in Business Analytics and Big Data, FGV (2019). Options, Futures and Other Financial Derivatives (FM360), London School of Economics, 2013.

Odebrecht Transport / Odebrecht Rodovias

Financial modeling and M&A for road concessions. Built and consolidated valuation and cash-flow models for infrastructure assets, including scenario and sensitivity analysis, to support the negotiation, purchase and sale of concession assets.

Dentsu Aegis Network / Isobar

Business analytics and financial modeling for technology projects. Designed data flows and processes and delivered BI reporting for major consumer brands.

Nivea / GoOn innovation hub

Led the data function of an innovation unit. Built cloud infrastructure and automated data processes, intelligence and reporting.

Nimbus, later Sellera

Joined as a consultant and became Head of Data. Built the platform's data and analytics layer: multi-tenant, parameterized reporting using Databricks for processing and Power BI Embedded for delivery.

HMData founder

Independent software and data engineering practice serving companies on Azure and Google infrastructure. Focus on custom applications, data infrastructure, process automation, AI-assisted workflows, and ongoing operation.

the work

Representative engagements, described without client names.

Across the practice, the role is consistent: an embedded data and software engineer who designs, builds and operates client data systems in dedicated, access-controlled environments.

Healthcare and clinical research

Business-intelligence platforms for a dermatology clinic and a medical-diagnostics company; a patient-experience data program in oncology; and a clinical-research data platform for a university group. These projects handle sensitive patient and personal data with access controls, encryption, and pseudonymization where appropriate.

Operations and back office

Workforce, scheduling and payment operations for a logistics operator; billing and time-tracking automation for a design studio; analytics for a property-management platform.

Analytics platforms

A proprietary out-of-home media analytics platform that consolidates data from multiple providers, applies a proprietary model to estimate campaign impact, and provides a dynamic application to simulate campaigns and standardize results.

Sports performance

A platform for a professional football club that consolidates athlete performance data from GPS tracking and from force and jump measurement systems into analysis for the coaching staff.

Certification and assessment

For a language-certification operation, a platform that internalizes and automates the processing of test results, keeping test-taker data pseudonymized and under strict privacy controls.

Common thread

Turning raw operational and personal data into reliable, automated systems and reporting, and more recently into AI-assisted workflows, while keeping that data protected and tightly access-controlled.

tools in practice

Tools used across real client projects.

A quick orientation to the environments Marcelo is used to building with, maintaining, integrating, or inheriting across client systems.

New apps and data workflows

New web apps usually center on TypeScript/JavaScript, Next.js/React, Vercel, GitHub and Azure SQL. Python, R, Databricks, Functions, Logic Apps, Storage and Key Vault appear in automation, pipelines, analysis or Azure operations.

Python SQL R TypeScript JavaScript Next.js React Vercel GitHub Azure SQL Databricks Azure Functions Container Apps Storage Accounts Logic Apps Key Vault

APIs and integrations

Geodata, vendor APIs, internal scripts, data pipelines, and controlled integrations between operational systems.

Google APIs Geodata Vendor APIs Internal scripts Data pipelines

Microsoft reporting and legacy tools

Still used where existing client processes depend on Microsoft reporting or low-code workflows.

Power BI Power BI Embedded Power Automate Power Apps

AI-assisted engineering

Part of the daily coding workflow for implementation, review, refactoring, documentation, and faster iteration.

Claude Code Codex
working style

Context-specific, controlled, and reviewable.

Data isolated and controlled

Each engagement runs in a dedicated, access-controlled environment. For privacy-critical work, data, credentials, code and documentation stay in infrastructure the client owns or controls.

Built around the client's context

Solutions are shaped to the client's real processes and people, not a product the operation has to bend around.

Clear, scoped integrations

Connections use documented vendor APIs wherever available, with the minimum access needed. When a standard API does not exist, any local adapter is explicit, scoped, reviewed and clearly labeled.

Clear confidentiality rules

NDA before access. No public names, no public references, no portfolio use without permission.

Transparent and reviewable

The client can see where data lives, which systems connect, what access exists and how the work operates. Documentation and operating procedures are kept clear enough for handover or independent review.

Continuity when useful

Most engagements are ongoing operation and evolution rather than build-and-leave. A defined, scoped project is also possible, with maintenance or continued improvement after delivery.

Private work, built to be understood.

Marcelo works almost entirely by referral and does not publish client names or use client work as marketing material. The emphasis is practical: systems that fit the real operation, protect the data involved, and remain clear enough to be reviewed, maintained, or handed over.