Professional Summary
Senior platform engineer who builds high-concurrency services and modern data platforms on AWS/Kubernetes. I design Python & Rust services orchestrated with Temporal and delivered via GitOps (ArgoCD, Crossplane, Helm), and I turn ML/AI POCs into reliable, revenue-impacting products. Highlights: scaled customer-facing APIs to thousands of RPS (+15% throughput), cut MTTR from ~3h to ~1h, reduced infra spend by 12–15%, and replaced a 3-day MapReduce/Postgres batch with a Kafka → Spark/Delta Lake pipeline that finishes in ~2h and powers <5-minute freshness in ClickHouse for interactive analytics.
Professional Experience
- Designed and automated AWS infrastructure with Terraform, enabling 100% reproducible deployments and faster service onboarding
- Developed internal Terraform modules to standardize observability and resource creation, embedding best practices across teams
- Built and operated high-concurrency platform services in Python & Rust, orchestrated with Temporal for retries, idempotency, and SLA-aware workflows
- Delivered GitOps-style dynamic infrastructure provisioning on Kubernetes using ArgoCD, Crossplane, and Helm
- Reduced MTTR from ~3h to ~1h via an observability stack (CloudWatch, OpenTelemetry, New Relic)
- Delivered 12–15% infrastructure cost savings via async compute, intelligent caching, and data deduplication
- Optimized database queries and execution plans for ETL pipelines, reducing processing time from hours to ~20 minutes
- Reduced database costs by migrating from RDS clusters to Aurora Serverless and introducing precaching strategies
- Led adoption of a PaaS-based UI regression tool, replacing manual Excel-driven checks with automated tests on every PR
- Drove a large-scale codebase refactor, eliminating ~80% of technical debt; reduced onboarding time from 4 weeks to 1 week
- Replaced a legacy MapReduce + Postgres batch system with Spark on Delta Lake fed by Kafka, cutting processing time from 3 days to ~2 hours
- Drove creation of the data lake (S3/Delta), authored Kafka producers for billions of scraped records
- Built real-time pipelines that reduced end-to-end data latency from hours to <5 minutes
- Deployed ClickHouse as the interactive analytics serving layer to power responsive, customer-facing UI exploration
- Designed and scaled a high-performance Django API, improving throughput by ~15%
- Collaborated with Data Science to design/optimize an NLP library, improving inference speed by ~20%
- Productionalized POCs and implemented custom features to automate manual processes
- Reduced cloud spend by ~$10k/month (~9%) by optimizing test strategies and lowering memory footprint
- Built and deployed a scalable training environment ensuring reliable performance across a wide range of LLMs
- Implemented comprehensive observability and hardened orchestration with retries, backoff, idempotency, and circuit breakers
- Built an event-driven ETL + status-tracking pipeline providing near-real-time visibility into task progress
- Cut storage costs ~20% via log lifecycle policies, selective version retention, and compression/archival
- Trained and coached non-technical personnel into full-time developers through structured curriculum
- Built and operated a mission-critical ETL pipeline processing millions of records/day, consolidating data from multiple POS systems
- Orchestrated Python workloads with Apache Airflow (NumPy, Pandas) and staged raw/curated datasets in a data lake (S3/Parquet)
- Served pre-aggregated views from the lake to accelerate category-level analytics and reduce warehouse load
- Deployed and managed production workloads on AWS (EC2, S3) for reliability and scale
- Achieved 100% test coverage with emphasis on high-value E2E tests, ensuring stability and minimizing regressions
- Built Spark-based reconciliation ETL for multimillion-dollar billing, comparing multiple sources to surface variances
- Operated and deployed services on Kubernetes/OpenShift with CI/CD, improving release reliability
- Shipped async APIs (Python, FastAPI, GraphQL, MongoDB) powering reporting dashboards
- Parallelized event submission to Kafka with multiprocessing + async I/O, delivering ~10× throughput improvements
- Led BI services for a major account; architected a scalable data pipeline processing hundreds of millions of transactions/day
- Built self-serve dashboards for non-technical users, reducing ad-hoc analysis requests and accelerating decision cycles
- Tuned SQL/NoSQL workloads (Hive, Oracle, MongoDB) to improve query performance and lower compute costs
- Coached junior analysts, raising team delivery quality and velocity
- Implemented Python-based forecasting (classification/regression) used in planning workflows, improving visibility into demand trends
- Delivered interactive dashboards for business stakeholders, increasing adoption of data-driven decisions
- Optimized SQL/NoSQL queries against large datasets (Hive, Oracle, MongoDB) to improve reliability and runtime
Technical Skills
Programming Languages
Python
Rust
JavaScript
Cloud & Infrastructure
AWS
Terraform
Kubernetes
ArgoCD
Crossplane
Helm
Orchestration & Workflows
Temporal
Apache Airflow
Data & Streaming
Apache Kafka
Apache Spark
Delta Lake
ClickHouse
Backend & APIs
Django
FastAPI
Flask
AsyncIO
Celery
Databases & Storage
PostgreSQL
Aurora Serverless
DynamoDB
Redis
MongoDB
Observability
OpenTelemetry
CloudWatch
New Relic
Languages
Spanish (Native)
English (Bilingual)
Education
GPA: 3.7/4.0 | San José, Costa Rica
Honor Graduate — Cum Laude Probatus, Software Engineering
Core Strengths
Professional Qualities
Constantly Curious
Great Communication
Team Player
Effective Planning