About Me
Started with BI dashboards in copper mining as a student intern. Now I co-author transformer research at IEEE and lead the engineering org of a payments platform.
I work across data engineering, ML, and team leadership. My background spans building production fraud detection on real transactional data and designing data warehouses that replaced months of legacy work — and I now lead the engineering org at a payments platform. I'm drawn to systems-level questions: how data flows shape decisions, how teams compound over time, what keeps a platform — software, data, and infrastructure — quietly reliable.
Key Achievements
Data Warehouse & ETL Architecture
Built a complete data warehouse with a new data model and end-to-end incremental loading — cut data loading time and costs by 85% while expanding the scope of data handled. The DW covers the main business data domains and consolidates information across different databases and platforms. I also built an ETL code generator for AWS Glue that standardized common patterns and accelerated new ETL development.
Business Intelligence Platform
Led the design and implementation of FlowPagos' BI platform — dashboards across the main business domains, several used daily in operations and decision-making meetings. Also built two meta-dashboards: one for data quality monitoring, another that tracks dashboard usage across the org to drive continuous improvement based on real use.
Machine Learning & AI
Shipped a real-time transactional fraud detection model based on anomaly detection — under 300ms response, end-to-end from experimentation to production (real-time and batch data processing, endpoint deployment). Built an MCC classifier to streamline merchant onboarding, later extended to classify SII codes (Chile) and SAT codes (Mexico). Designed a multi-agent architecture for an enterprise customer-facing chatbot — sub-agents specialized by task domain.
MLOps & Development Standards
Built a standard repository for AI projects: GitLab CI/CD with Jira approval gates, quality tests, static analysis, and automated AWS deploys via CDK. Includes reference scaffolds — SageMaker pipelines, FastAPI/SageMaker endpoints, EKS deployment, pre-commit checks, and infrastructure as code — plus a LangChain LLM template. Standardizes the team's path from prototype to production.
Team & Area Leadership
As Data & Analytics Lead, led a team of 4 across data engineering, BI, and AI projects. Set up recurring team rituals that built cohesion, ran a standardized quarterly feedback process, and led hiring. Also built Jira automations using LLMs — auto-generating standardized descriptions for different requirement types — to keep project management consistent across the team.
My Approach
I work problem-first, not tool-first. The choice of stack, model, or pattern is downstream of understanding what's actually broken and which constraint matters most. The same applies to technical and people problems alike: whether it's an ETL bottleneck or a team's quarterly priorities, I diagnose before proposing solutions.
What Drives Me
What drives me is depth and breadth at the same time. The year I'm tuning a transformer for fraud detection is also the year I'm reorganizing how a team prioritizes its quarter. The two muscles feed each other: engineering taste sharpens leadership decisions, and operational reality sharpens the technical ones.
IEEE CAI 2025 Experience
With Vasco Cortez I worked on a modification to positional encoding in transformers — adapting it to represent the actual temporal distance between transactions in a sequence, rather than just their order. We tested it on transactional fraud detection. The paper was accepted at IEEE CAI 2025 and I presented it in Silicon Valley. Beyond the talk itself, the most valuable part was the conversations with other researchers and practitioners in the field.
Read paper on IEEE XploreUniversity Initiatives
DataLab Co-founder
Together with Diego Quezada, a classmate from ML courses, we created a proposal for a Data Science and AI laboratory for the university. We laid the foundations for DataLab and secured funding from the UTFSM-Scotiabank alliance, which subsequently allowed the implementation of a laboratory at both Campus Casa Central Valparaíso and Campus San Joaquín.
1st Place — Software Exposition
Led a team to 1st place out of 10 teams at UTFSM's XXIX Software Exposition (annual university software competition) — judged on build, technical depth, and presentation.
Research Assistant
I was invited to participate in a project to improve the Computer Science Engineering program, where we conducted interviews with students and professors, analyzed results, and produced conclusions and recommendations that were later used to improve the program.
IT Project Management Diploma
Thanks to a scholarship, I completed a diploma in IT Project Management, which allowed me to enter the workforce with a project and team management perspective that complemented my technical interests very well.