Solution Architect and Developer in Santo Domingo, Distrito Nacional, Dominican Republic

I refactored the pipeline structures for Zebra, a Rust-based Zcash Node.js reimplementation, using GitHub Actions and Google Cloud Platform as core technologies to automate the entire CI/CD. This helped the core engineering team focus on development, ensuring bugs were not merged into the main system while the queuing system was also implemented.

Build times were reduced from three hours to 50 minutes without cache and five seconds with cache, improving the merging speed by more than 300% when combined with the automated queue and merging system.

Led the fast-paced countrywide implementation of the software and infrastructure to tackle the logistics and vaccination needs for COVID-19 using Odoo framework and Google Cloud Platform with Kubernetes to accomplish a fast and automated solution.

This project required taking a whole infrastructure, which had two years working in OVH using Proxmox with ZFS and automatic service discovery (Nginx with DNS) and over 100+ LXC servers, each with their application (Odoo) and database (PostgreSQL).

This scenario was causing an exponential overhead when the application had to be updated, as each server needed manual maintenance, with a 6+ hours maintenance window.

The first step to solving this was to accomplish a more stable and standardized solution. Dockerizing the application assured having a base image known by the sysadmins and developers, which could be used in multiple environments (development, staging, and production). This image was open-sourced to be used by the global Odoo community.

As this project required a more "explicit" infrastructure definition, Kubernetes was the chosen tool to which to migrate. The infrastructure was deployed with automated SSL certificates, Istio, and Envoy for networking, observability, networking, and special configurations for caching and faster TTFB. Some clients had web pages that needed low response times for SEO.

This client had to wait for months to visualize the work, even if it was a WIP, as the development team had to provision servers and do time-consuming tasks, which led to needless delays.

Continuous integration and continuous delivery led to the inclusion of DevOps practices in the institution, guiding the team with Docker use and Dockerfile best practices, pipeline design with Github Actions, solutions architecture, and code standardization, linters, pre-committers, security scanners, etc. The development lifecycle was supported by Cloud Run on most products, from building development instances before merging to master, generating staging instances as a pre-production stage, and deploying to production using Github Releases, all fully adapted to the team size, actual knowledge, and capacity.

Deliveries are now being done under four weeks, and tech leads can see and test the front end, back ends, APIs, etc., as soon as the developer working on the system, or a new feature, makes a PR to a main/master branch, creating a mirrored environment (platform) from production.