Peter Bui 🔗 Email & Phone Number

Alameda, Ca, Us

San Francisco, California, Us

San Francisco, Ca, Us

PlanGrid was acquired by Autodesk in 2019* Started as the engineer overseeing 20 datastores in a single US region. Grew the team to 8 engineers that managed 200+ datastores (RDS Postgres, AWS Elasticache Redis, AWS Elastisearch) across four different regions (Australia, EMEA, 2x US).* Led effort to migrate two large monolith service databases (4TB Postgres, 2TB Mongo database) to RDS Aurora and MongoDB Atlas, respectively, and designed the migrations to fit within a two-hour downtime window for each database. As a result, the effort reduced RTO from 2 hours to 20 minutes for both databases and reclaimed 8-weeks of person-hours per year for maintenance.* Designed the migration strategy to migrate old Cloudformation-managed Kubernetes applications to Terraform-managed AWS EKS clusters with zero downtime.* Promoted and designed API interfaces for inter-team communication to move away from JIRA ticket-based requests.* Developed a project execution process that leads to delivering 80% of projects on time for four quarters in a row. Evangelized, trained, and standardized the process with two peer teams resulting in similar delivery rates.* Reviewed, coached, and set a standard for engineering change proposals within the organization.* Continuously educated/trained engineering teams on database self-service tools and database management practices to increase development team velocity when building features.

San Francisco, Ca, Us

As a tech lead: * Led a year-long project to provision a second region for PlanGrid products in a repeatable manner with three teams totaling 15 engineers. The system used Terraform/Terragrunt to provision a new region from scratch with a VPC, Kubernetes clusters, Spinnaker, and 40+ AWS datastores (Postgres/Redis) in two weeks. This system has since provisioned three additional regions since inception driving revenue for ACS that was not possible.* Learned project leading, starting with a single project with two developers. By the end of my tech lead tenure, I ran 2-3 simultaneous projects at a time with 2-3 developers per project that exected between 2-3 months.* Redesigned and migrated bespoke 3TB Postgres cluster architecture to enable a reduction of RTO from 48 hours to 2 hours and RPO from 24 hours to a few seconds.* Took on a daunting task to develop the database strategy for the engineering organization that would lead/forced engineering teams to move towards a service-oriented architecture and enable multi-region deployments.* Wrote and utilized written engineering proposals to persuade, cooperate, and lead design changes to the distributed systems across the team and within the engineering organization.* Developed a standard postgres configuration based on prior incidents for all Postgres databases enabling faster feedback loops and removing long-running queries from consuming all database resources.

San Francisco, Ca, Us

* Replaced critical EC2 scaling system with more performant auto-scaling system built with Cloudformation, EC2 Auto Scaling Groups, SaltStack, Docker containers, and a golang command-line deployment tool. As a result, it reduced critical asynchronous job processing time by 95% and reduced incidents by 50%.* Migrated PlanGrid's compute infrastructure from Heroku to AWS, Kubernetes, and Spinnaker with zero downtime and paved the way for achieving SOC2 Type II certification.* Replaced bespoke manual/Ansible-based configuration system with a highly available setup for SaltStack Masters that manages all on-instance configuration for every EC2 instance in PlanGrid's AWS infrastructure.* Studied and developed Infrastructure as Code techniques with Cloudformation and git to manage PlanGrid AWS cloud infrastructure. In addition, I promoted the methodology to the rest of the team. As a result, we replaced 95% of AWS resources in staging and production environments with infrastructure as code in two years.* Stepped up to help educate and standardize the method of releasing schema changes to our monolithic 2TB postgres database for 20 backend engineers.

San Francisco, California, Us

I wore many hats: frontend, backend, infrastructure.* Replaced Solr-based search faceted browsing system with Redis sorted sets that covered 95% of the original feature set but 35% increase in performance and scalability.* Rebuilt the site's music player MySQL backend with a combination of MySQL and Redis that enabled TTL-based records without requiring DELETE queries from slowing down the performance of the database.* Maintained a 99th percentile database latency of 10ms for queries for the central Ruby on Rails application with aggressive caching, query optimization, and schema denormalization.* Replaced a ruby-based, async job, image resizer with golang-based, docker containerized, dynamic HTTP resizer to increase the speed at which the development team could release new designs to meet advertisement demands.* Built and customized a Varnish load balancer to handle spikes in traffic due to Justin Bieber fans swarming the site.* Built metrics observability with Graphite and statsd to enable better debugging of our distributed system.* Managed the AWS cloud infrastructure using Chef and Knife

San Francisco, Ca, Us

* Backend programmer focusing on adding features to help the sales team manage our clients* Refactor sign up system code to reduce complexity and add robustness

Us

* Developed modularized widget system to allow customization of sites on the company's community portal platform * Implemented Tickle.com's test taking functionality into community portal platform for the Affinity Labs takeover of Tickle.com * Upgraded community portal platform's legacy HTML/CSS/Javascript to the latest web standards to ease development of new features * Designed and implemented a multi-page, sign up flow to allow the marketing team to customize lead-gen sign up process -- previously, the sign up process consisted of a static four page flow with no user customizations * Led interface redesign of lead-gen application's admin interface to ease management of Affinity Labs' 100+ lead-gen sites * Developed domain squatting application for Monster.com's 5,000+ unused domains* Redesigned legacy ad selection system to utilize Redis instead of MySQL to integrate with Monster.com's interstitial adverts

* Evolved engineering process by involving upper management in weekly iteration meetings -- this allowed management to understand scope as well as help them properly plan around engineering releases* Led the transformation of the current application into a REST based web service for internal application use * Built a basic data warehousing application with Rails to analyze fan club sales data

* Led legacy, fan club application rewrite with Ruby on Rails * Introduced test driven development, unit testing, and code refactoring to reduce defects * Enhanced Rails' test fixtures to allow them to load in multiple databases * Developed Rails MySQL adapter to split write and read queries to master and slave database servers, respectively * Helped the engineering manager develop a feature tracking system to reduce 2-3 month release cycles down to one month and allowed long term planning for Sparkart's executive team -- (think of a primitive Pivotal Tracker application) * Created application deployment scripts for 15+ servers (application/database/asset servers) * Introduced pair programming into the software engineering process to encourage peer review and higher quality code