Senior Data Engineer · A decade at MercadoLibre
In 2025 I built a governed data platform from zero — 203 production tables, 93 scheduled jobs, and six business domains migrated to a Bronze/Silver/Gold mesh. On weekends, I decode the emotional arcs of Studio Ghibli films.
I'm Eduardo Franklin — a Senior Data Engineer who turns messy data into clear decisions. I've spent a decade at MercadoLibre, and I've been in tech since 2008 — back when the job meant running cables through ceiling tiles in a server room.
That long road taught me what the CS-degree shortcut often skips: great data engineering isn't about pipelines, it's about understanding the business problem first, then building the simplest solution that scales.
Currently, after working with seller invoices, logistics, and loyalty teams across multiple projects, I'm helping the Tech Corporate business unit build a strong data foundation from a legacy sandbox after years of rapid growth in capex and opex.
When I'm not working, I'm practicing calisthenics, climbing, hiking, biking, or eating sweet treats with my elder dog.
No CS degree, no bootcamp, no neat junior → senior ladder. Just a long road from running cables to building data platforms — and the breadth that came with it.
2008
I started in IT infrastructure and service desk — installing operating systems, running network cable, debugging hardware (and people). The classic unglamorous entry point, and the best training in asking the right questions I ever got.
2015
A week after I joined KPL, MercadoLibre acquired it — and overnight I was part of Latin America's largest tech company. The transition to data wasn't instant; I kept gravitating toward logs, queries, and dashboards.
2018
I started building KPI visualizations for my team in Mercado Envíos (Logistics). Nothing fancy — just charts that helped people see what was happening. They got noticed, and data became the job for good.
2024
I took ownership of a corporate-tech business unit's data foundation: a legacy sandbox, a dozen ungoverned dashboards, a BigQuery environment no one fully understood, and a small team looking for direction.
2025
Built a governed data mesh from an empty environment — 203 production tables, 93 scheduled jobs, six business domains migrated to a Bronze/Silver/Gold architecture. Read the case study →
Grouped by how I actually use them, with years of hands-on experience — and a link to proof for each. Core = daily, deep; Proficient = comfortable in production; Working = shipped real work, still growing.
The work behind the numbers — problem, architecture, and results.
Senior Data Engineer · MercadoLibre (Tech Corporate BU) · 2025
Turning an empty, governed BigQuery environment into a production data platform — 203 tables, 93 scheduled jobs, and six business domains migrated to a Bronze/Silver/Gold architecture in a single year.
Data & Analytics Engineer — Mercado Envíos (Logistics), MercadoLibre · 2018–2024
Six years embedded in logistics: first the dashboards that made invoicing times and platform adoption visible for product and operations, then the machine-learning models that grew fulfillment adoption and NPS while cutting churn.
The emotions of Studio Ghibli, decoded
A data engineering passion project that analyzes the emotional arcs of Studio Ghibli films using NLP and machine learning.
From My Neighbor Totoro's gentle wonder to Princess Mononoke's fierce beauty — every film tells an emotional story that data can reveal.
Technical deep-dives and project breakdowns
AI started as a way to build a single pipeline faster. It became a way to scale a whole team — encoding the standards that used to live only in the senior engineer's head into tools anyone can use.
At the start of 2025, our Data Mesh had 5 jobs, 5 tables, and a handful of daily users. By mid-2026: 150 jobs, 288 tables, 90 average daily users (up from 12), and a steadily growing compute footprint. Here's the full growth curve, with charts.
A Data Mesh with no operating manual is a platform only its engineers can use. Here's what we shipped to make ours serviceable: a formal RACI, a two-track access catalog, ten documented governance processes, and the pipeline service catalog that replaced the Slack messages.