Javier
Aragón Navarro

Business Intelligence & Analytics Lead — 8+ years turning complex operational data into measurable business outcomes through data engineering, analytics, and applied AI.

javier@horizonlabs.work +1 226 759 8674 Windsor, ON · Canada
Get in Touch View Experience Resume & Docs
8+
Years in Data
500+
Daily Decision Makers Served
400+
Hours Freed Annually
65%
Dashboard Adoption Lift
5
Countries at Scale
99.9%
AI Agent Metric Reproducibility

Foundations first. AI on top.

Because a number you can't trust, can't explain, or can't act on isn't an answer. It's decoration.

Open to Work — Always interested in meaningful conversations about Data, BI & AI innovation.
Proof

The thesis, in production

Two systems I built end to end, where trusting the number, knowing what it meant, and acting on it weren't optional. Plus the essay that argues why they're built this way.

Case StudyE-commerce · Amazon FBA2026
Clarity for an Amazon seller: from blind spots to per-product decisions
An operational-intelligence platform that allocates Amazon's bulk costs down to each product, flags which ones quietly bleed margin, and answers restock questions in plain language.
View case study
Case StudyEnterprise ops · Logistics2026
Just ask. Get a number you can trust.
A conversational analytics platform that turns plain-language questions into audited, governed queries and causal diagnosis — for operators and managers across five countries, not just analysts.
View case study
EssayBI ArchitectureMay 2026
BI's Future Isn't a Dashboard With AI On Top
The long-form version of the thesis: why modern analytic products need foundations, not retrofits, and an architecture that separates the permanent investment from the consumable.
Read essay

Beyond full-time roles, I take on select data & AI projects through HorizonLabs. See services →

The track record

Where it came from

Eight years building toward that thesis — from logistics intelligence at Latin American scale to AI-grounded reporting and financial analytics.

Semi-Senior Data Analyst — BI Operational Excellence (Regional)
Jul 2025 – Present
Lead analytics strategy for operational KPI alignment, adoption measurement, and performance tracking across Mercado Libre's logistics & fulfillment operations in Brazil, Mexico, Argentina, Chile, and Colombia.
Mercado Libre · Remote (Windsor, ON)
Mercado Libre (NASDAQ: MELI): The Amazon & eBay of Latin America — operates in 18 countries with $28.9B annual revenue (2025). Leading e-commerce, fintech, and logistics player. Publicly traded with all-time highs near $2,600+, rated "Strong Buy" by 17+ analysts. One of the most valuable tech companies in the region.
  • Brought to production the architecture from my essay "BI's Future Isn't a Dashboard With AI On Top": built a governed conversational BI system (deployed as a Claude plugin) — an operational codex of business rules feeds a canonical-metrics layer and a nested SQL-builder + audit-agent pipeline connected to BigQuery via MCP; validated in beta at 99.9% metric reproducibility vs. the official reporting product, grounding every output to flag hallucinated answers from other AI agents
  • Architected a modular Labor Management System in Tableau measuring hourly workforce productivity — scaled across Brazil, Mexico, Argentina, Chile & Colombia, reaching 500+ daily decision-makers
  • Built a causal productivity-analysis system diagnosing cost-per-package deviations in cross-docking operations: an anti-leakage feature architecture eliminating algebraic confounding, three-tier model pipeline (OLS decomposition + causal OLS selection + XGBoost + SHAP with time-series cross-validation), structural-break reconstruction and automatic lag detection; served as a FastAPI REST service for both workflows and LLM agents; reduced diagnostic time from hours/days to seconds at the pilot facility in Mexico
  • Delivered an Operational Management Monitor unifying workforce, transportation & fulfillment data streams, freeing 400+ hours annually in Audit and Operations
  • Led KPI governance across 5 countries: 15+ dashboards, 100+ audited metrics, driving a 65% increase in official dashboard adoption
  • Led pipeline migration & validation of the regional adoption-metrics product (cross-validated 299 calculated fields, 49+ filters) and built an automated documentation framework converting 49 Tableau workbooks into JSON-structured playbooks, eliminating a manual documentation bottleneck
BigQueryTableauPythonXGBoostSHAPFastAPIMCPn8nLLM Agents
Founder & Principal Data Consultant — HorizonLabs
Jan 2022 – Present
Windsor Machine Group & Just Toys International LLC · Canada / USA
Partner with companies to design and implement end-to-end analytical products — from ETL automation and data pipelines to BI dashboards and technology adoption strategies. Every solution anchored to business outcomes and scalability.
  • Architected Power BI financial reporting layer for Windsor Machine Group (Canadian precision manufacturing — auto parts): consolidated 8+ Excel sources into unified data model; implemented WoW/MoM variance tracking and exception alerting; reduced reporting prep time from 3 days to 2 hours per cycle
  • Shipped an operational-intelligence platform for Amazon FBA (June 2026) for Just Toys: Airflow/pandas pipelines ingest raw Amazon reports (settlements, inventory, fees, advertising) into PostgreSQL; computes true SKU-level monthly P&L with cost of goods; scores product health via ML benchmarks + Prophet forecasting; generates a plain-language verdict per product (restock / reprice / liquidate / optimize) via LangChain + Claude
  • Exposed the platform through a custom MCP server (FastAPI, Docker, Cloudflare, Google SSO) — non-technical stakeholders ask "what should I restock?" in natural language and receive pre-computed, auditable answers; includes a Next.js operational dashboard
  • Foundation work: full-stack FBA profitability system reconciling bank statements to SKU-level P&L with waterfall analysis in Looker Studio; rentability KPI framework (CM1–CM4, ACOS, storage, advertising) pinpointing pricing, catalog, and inventory issues
PythonPostgreSQLAirflowFastAPIMCPProphetLangChainNext.jsDockerPower BILooker Studio
Semi-Senior Data Analyst — Line Haul Operations
Jan 2024 – Jul 2025
Mercado Libre · Mexico City
Owned regional planning and standardization strategy for Mercado Libre's line haul network across Mexico. Responsible for collecting operational KPIs, proposing compliance strategies, validating operational standards, and enabling on-site teams with verified data sources.
  • Owned regional planning & standard adoption strategy for line haul network (Mexico): proposed compliance strategies, validated operational standards, enabled on-site teams with verified data sources; reduced data discrepancies by 65%
  • Built BigQuery pipelines & executive dashboards measuring KPIs, route compliance, deviation patterns across logistics network; enabled root-cause analysis for underperforming facilities
  • Developed ML models (Linear Regression, Decision Tree, Random Forest) for shipment demand forecasting and capacity planning; improved facility-level resource allocation accuracy by 18%+
  • Engineered linear algebra optimization solver for MLP (Modal Load Planning) allocation, balancing cost and operational constraints across regional hubs
BigQueryscikit-learnPythonSQLLinear OptimizationTableau
Data Engineer — Customer Behavior Analysis
Oct 2022 – Jan 2024
Bancoppel · Mexico City
Bancoppel: Unique Mexican conglomerate operating 3 verticals simultaneously — banking, retail, and e-commerce. Total assets: 175.3 billion Mexican pesos (~$9.3B USD) with 11% YoY growth. Credit portfolio: 70.3B pesos (17% growth). Rare integrated business model in LATAM — one of few institutions combining financial services, retail operations, and digital commerce.
Led customer behavior analysis and profiling for Bancoppel's banking, retail, and e-commerce operations. Collaborated closely with Sales and Marketing to build horizontal customer segmentation models driving strategic decisions.
  • Led customer behavior analysis & profiling across banking, retail, and e-commerce (Mexico & Argentina); categorized consumption profiles into 12+ segments for Sales & Marketing-driven campaigns
  • Architected 360-degree customer profiles combining transaction history, product affinity, and payment behavior; delivered to Sales & Marketing for campaign segmentation and retention programs
  • Developed & maintained 150+ audited KPIs across data marts; automated compliance reporting and variance detection; reduced metric errors by 40%
  • Engineered Python-based ETL workflows for daily ingestion of cross-platform data; implemented idempotent logic and quality checks ensuring reliability for downstream analytics
PythonSQLETLData ModelingBigQueryTableau
Commercial Operations & Analysis Specialist
Apr 2017 – Oct 2022
Just Toys International · Mexico City
Just Toys International: Hasbro competitor in the Latin American licensed toys market. Operates across multiple countries with strong retail distribution in the toy & entertainment space.
Cross-functional operator bridging retail sales, supply chain, and manufacturing. Drove data-informed decision-making across the entire value chain — from daily sales monitoring to months-ahead demand planning with Asian factories.
  • Retail sales tracking & analysis: monitored daily sell-through, SKU performance, and retail compliance across major US retailers; escalated trends within 24 hours for corrective action
  • Forecasted seasonal demand patterns with 90%+ accuracy; coordinated production schedules with factories (China/Vietnam); reduced peak production costs by 25–30% through anticipatory scheduling
  • Led supply chain crisis management during 2021 shipping crisis: coordinated container consolidation strategy; protected delivery timelines and retailer compliance through proactive rerouting
  • Built proactive workforce forecasts tied to seasonal production ramps; enabled hiring before demand spikes; achieved consistent positive compliance with major retail partners
Excel AdvancedFinancial ModelingSupply Chain AnalyticsForecast ModelsData-Driven Strategy
Technical Stack

Skills & Tools

Languages
SQL Python R DAX Power Query (M) VBA Bash
BI & Visualization
Power BI Tableau Looker Studio Semantic Models Workspace Admin
Cloud & Data Platforms
Google BigQuery PostgreSQL GCP Docker Cloudflare Microsoft Fabric Azure
AI & Machine Learning
LLM Integration Multi-Agent Orchestration MCP Server Development Prophet Prompt Engineering RAG / Q&A Bots XGBoost SHAP statsmodels scikit-learn Causal Analysis Random Forest Forecasting Clustering
Orchestration & Automation
Apache Airflow dbt N8N FastAPI REST APIs ETL Design Cron Jobs
Methodology
Lean Six Sigma KPI Framework Design Data Governance SDLC Stakeholder Mgmt
Credentials

Certifications

🤖
Claude Code in Action
Anthropic
March 2026
🚚
Supply Chain Analytics Specialization
Rutgers University / Coursera
September 2022
🔗
LangChain V1 Essentials — Python
LangChain
April 2026
📊
Google Data Analytics Professional
Google
📈
Complete Power BI Desktop Course
Udemy
⚡
Lean Six Sigma Yellow Belt
IASSC
⚙️
Apache Airflow & dbt
In Progress
In Progress · 2026
Education

Academic Background

Rigorous quantitative foundation from one of Latin America's top research universities.

Master of Science — Earth Sciences
UNAM — National Autonomous University of Mexico
2014 – 2017
Honorable Mention
Bachelor of Science — Physics
UNAM — National Autonomous University of Mexico
2007 – 2012
Contact

Let's work together

Available for full-time roles in Windsor–Leamington and surrounding areas. Open Work Permit — no sponsorship required.

✉️
Email
javier@horizonlabs.work
📞
Phone
+1 226 759 8674
📍
Location
Windsor, ON · Canada
🛂
Work Status
Open Work Permit