Gabriel Florio
I really, really like computers. Also metaphysics.
Experience Summary
Once you are done with this summary, you can find out more about me in my resume.
Table of Contents
Expert
Docker
I have been using Docker in production since 2017. I think it’s a superb tool, even though sometimes it does have a few counter-intuitive and hard to debug problems.
I’m also comfortable with image optimization and multi-stage builds.
See also: DevOps.
Google Cloud Platform (GCP)
I have been using GCP since 2017. I think it’s a great platform for running all kinds of workloads, and I’m comfortable with most of its characteristics and quirks. I have done official Google courses to improve my knowledge of the platform, and I have good experience with GKE, GCE, GCS, Cloud SQL, Pub/Sub, Dataproc and many other services.
See also: DevOps.
Helm
I have been using Helm daily since 2020. I think it’s a great tool for managing Kubernetes deployments, and I’m comfortable with writing complex charts from scratch, including subcharts and dependencies. At Roon Labs, I used a combination of base charts and CLI tools to design a new developer experience with deployments, making it easy for developers to follow best practices, and for SRE enginners to roll out changes to multiple services at once.
Indoor Positioning
From 2015 to 2017 I have worked mostly on indoor positioning. During that time, while the focus was on using commodity hardware, I have tried just about any technology you have heard about (and probably some you haven’t).
I have used machine learning methods extensively, as well as methods based on wave propagation (RSSI, AoA and TDoA), complemented with dead-reckoning. Ended up with an approach that could combine multiple results using Bayesian sensor fusion.
NodeJS
I have been using NodeJS since 2015. It’s a great platform for building lightweight containers and I’ve used it a lot to build smaller, functional-style, workers that would consume events from RabbitMQ or batch process data.
In my opinion, its biggest downside is Javascript itself, being a language that does not lend itself for building large systems with heterogeneous teams. TypeScript is a step in the right direction to alleviate that. Remains to be seen if it will become a reliable standard.
In April 2020 I was verified as a top 5% NodeJS Expert by PluralSight IQ.
Python
I’ve been using Python personally since 2009, and professionally since 2015. I think it’s a great language for a lot of tasks, and the ecosystem has some of the best tools around for web services, machine learning and scientific computing in general.
The main downside, which generally is performance, can normally be mitigated by careful profiling and building an extension in Cython. My personal project pytimeset was inspired by one of these occasions. By using it, I have achieved a run time reduction of around 85% for a fairly complex algorithm I had in production, which was taking hours to run for some of our bigger clients.
I believe I am able to write Python in a very pythonic and clean way, and I follow PEP8 rigorously. In April 2020, I was verified as a top 1% Python expert by PluralSight IQ.
REST
A lot of people use “RESTful API” to mean simply a JSON API over HTTP, but I think there’s really a lot of concrete gain to be had by adopting REST more closely to what Roy Fieldings first envisioned.
I have been using REST more rigorously 2017, although I have been designing “RESTful” JSON APIs over HTTP since 2015. I have good experience on properly using the HTTP verbs (including the often forgotten OPTIONS), resources, and HATEOAS.
Software Architecture
From working under a software architecture professor, who then became more or less my mentor for the whole college course, I got a strong theoretical basis in the field. I feel comfortable with viewpoint modeling and with applying reference architectures to concrete problems.
That said, I know most people aren’t like that, and the word itself has a (probably deserved) stigma around it, so I try to have a very lightweight approach to architecture in practice. I believe that we can get the whole team to think architecturally just by asking the right questions at the right stage of a project and working together to answer them, and that we can get a lot of concrete value from doing that. No UML required, and definitely no Big Design Upfront.
Experienced
Agile
I strongly believe “agile” is an adjective, not a noun. Therefore, agility should be built in on everything we do. When building a successful startup from scratch in a economically unstable country such as Brazil, agility was vital to the survival of the company. I think that helped me build a natural agile mindset.
Algorithms
I’m a bit of a Logic nerd, therefore I’m also a bit of a algorithm nerd. I feel comfortable analyzing, designing and tweaking many kinds of algorithms, but I especially like heuristic searches. In fact, tweaking algorithms is pretty much all I do in my pet project OperaPlan.
ArgoCD
See GitOps.
Artificial Intelligence
Being the algorithm nerd I am, many areas of AI interest me:
- I have built, deployed and maintained machine learning models when working on the Indoor Positioning project.
- I also have built, deployed and maintained heuristic search algorithms on the Process Optimization project
- I have a solid grasp of Game Theory. I’ve been wanting to get into Agent-based Computational Economics, but still haven’t really gotten around to that yet.
See also: Machine Learning, Optimization.
Debian
See Linux.
DevOps
I think DevOps as the name of a position or a team is a bit weird, but I get what you mean. In that sense, I think I was lucky, because up until now I have always worked in places with a strong DevOps culture. I have always been responsible for taking care of my code during all its lifecycle, including running and monitoring it, and I learned a lot because of this.
Later on, I had the opportunity to help rebuild from the ground up much of the infrastructure at Roon Labs, which was a great learning experience working with DevOps and GitOps at scale.
Django
I first had contact with Django in 2014 and it was love at first sight. I have chosen to use Django + Django Rest Framework for every application I’ve written in Python since 2016. I know pretty well the ins and outs of the framework and I think it’s a great tool for many kinds of projects.
I have written applications with it in the “default” architecture as well as in conjunction with an “Hexagonal” architecture, in order to make the domain model stand out better, unhindered by infrastructure code.
See also: Python.
Domain-Driven Design
Essential technique in my toolset. I have been using DDD and pushing it in my teams in one way or another since 2017. Even though the beginning was a little awkward (the learning curve can be pretty steep), I now consider myself able to properly apply both the tactical and strategic patterns in a way that makes sense to the project I’m working on.
Even though some of the techniques aren’t really well suited for all software projects, I believe that a “domain first” approach is always beneficial.
Fast paced environments
I can’t think of a faster environment than a tiny startup rising up from the ground. Honestly, I love this kind of environment (in fact, our office was on a old house where two of the team members lived, so there was work going on all the time. I have slept on their couch more than once). I work my best when I can go to sleep with a problem in my head, wake up with the solution, share it with everyone and go make it.
On the other hand, such environments are prone to the “permanent sense of impending doom syndrome”, and it’s not a easy task for management to avoid that. As such, I think that a open feedback culture is a hard requisite for having a healthy fast paced work environment. Sometimes we just need to take a step back and take a good look at our priorities, everyone works better when they’re feeling safe.
Flask
I have used Flask on a few projects since 2016. I tend to prefer Flask over Django when using NoSQL databases.
Git
I have been using git professionally almost every day since 2017. At this point, it’s almost second nature for me. I have worked with trunk based development and with feature branches workflows.
GitOps
I’m a strong proponent of GitOps. I have been using it since 2018, in different forms and I think it’s a great way to manage infrastructure for many kinds of projects. The specific tools will vary by project, but the benefits of having a detailed and precise log of all changes to the infrastructure are too good to pass up.
Java
I first learned Java when I was 13 years old, because I wanted to make cross-platform GUI programs for my friends. Since then, I have been using it on some of my projects (mostly whatever needs high throughput), but more intensely since 2018.
Java doesn’t get much love out there, but it’s one of my favorite languages. It’s compatible with absolutely everything, the JVM is a wonder of modern engineering, and the language lends itself very well for work with larger, heterogeneous teams.
I think most of the hate on Java is caused by people misusing or abusing OOP. I have good OOP foundations and practice, I keep myself updated on new Java features, and my copy of Effective Java always close by.
Kubernetes
I have been using Kubernetes since 2017 for managing several projects in a auto-scalable cloud of around 20 nodes, serving upwards of 2000 requests per second, including many mission-critical services. I have experience configuring clusters from scratch using Terraform, GKE and Helm, and also applying the current best practices for availability and security.
See also: DevOps.
Linux
I have been using Linux on my personal computer since 2008, and then on almost every server, container and VM I have ever managed. I would say I have a pretty good grasp on Linux operation and administration. Through the years, I have used Ubuntu, Debian, Fedora and Alpine.
See also: DevOps.
Machine Learning
I first started looking into machine learning when I started the Indoor Positioning project in 2015. Thankfully, I had a solid statistical background from engineering school, so many of the concepts were relatively easy to grasp.
Because of that, I have acquired experience in building, deploying and maintaining machine learning models in production, complete with unsupervised outlier detection, data cleaning and normalization, cross-validation and metaparemeter optimization by local search.
As for the actual algorithms, I tend to prefer something tweakable and interpretable than the absolute best performance, because I think that’s what brings more value to most real business cases.
Microservices Architecture
Back in 2015, I had the opportunity to participate in a completely greenfield project, and we chose the way of microservices. We did a pretty good job, I believe, because the services were really autonomous, the communication was entirely done with asynchronous messages, and we did eventual consistency right.
Even so, I probably wouldn’t do that again. Experience has nudged me in the way of a little more coarse autonomous services, as aligned as possible with the subdomains of the business. There is some discussion to if that qualifies as “micro” services.
Then again, I don’t think “microservices” really describes an architecture, it’s more like a deployment strategy. While that’s important, it’s only a part of the architecture.
See also: RabbitMQ, Software Architecture
MongoDB
I have been using MongoDB on some of my projects since 2016, sometimes as an auxiliary database, sometimes as the main database. Although it’s usually not my first choice (too easy to run on nasty N+1 problems), I think MongoDB can work very well when paired with Domain-Driven Design, as properties of a Document fit quite nicely those of an Aggregate.
MySQL
See SQL.
Optimization
Sort of my favorite theme in Computer Science right now. I think that applications based on heuristic optimization have a potential to transform a lot of businesses, especially when paired with machine learning.
My graduation project in college was about this and, since then, the more I learn, the more excited I get. On my job, I have designed, implemented and deployed a optimization system for job distribution among maintenance workers of a large industrial client, complete with a constructive heuristic and local search.
I also have a pet project that is really just an excuse to tweak around algorithms and see what happens.
PostgreSQL
I have been working with PostgreSQL since 2016, and managing it in production since 2020. I have a good grasp of the best practices and the most common pitfalls related to operating simple databases.
See also: SQL.
RabbitMQ
It’s a great tool to build autonomous services around. I have been using it in production since 2016 for processing hundreds of messages per second and I’ve found it’s really reliable and relative easy to manage.
I have also used it just for scheduling asynchronous tasks (such as with Celery), but I think it might be a little overkill just for that.
Redis
I have been using Redis since 2016, mostly as a cache. I’m also used to managing it in production, both standalone and inside Kubernetes clusters.
Remote working
I have been working with varying degrees of “remoteness” since 2015. The best thing about it, in my opinion, is being able to work with the best people from all over the world, without being restricted by geographic barriers.
Spring Boot
My framework of choice when building Java applications. I have been using it professionally since 2018, and I’m always amazed at the sheer amount of functionality it offers, and the quality of (most) design decisions that went into it.
Of course, because of that, one risks having the whole application swallowed up by it, but that can be avoided by having a separate domain model, independent from the framework.
See also: Java.
SQL
I have been working with SQL databases since 2015. During this time, I have worked with Microsoft SQL Server, MySQL and PostgreSQL. I have used ORMs (Django and JPA2/Hibernate) and have experience hand-crafting optimized queries.
As far as building and maintaining an application goes, I would say that I have no problems using any of those tools whatsoever.
Terraform
I have experience using Terraform to manage infrastructure as code, such as Kubernetes clusters, databases, load balancers and other cloud resources. I have written custom reusable modules, and have experience managing infrastructure in an immutable paradigm with it.
Ubuntu
See Linux.
Some Experience
Ansible
I’ve used Ansible for about two years, to manage both on-premises and virtual cloud servers. In 2017 my team has moved on to cloud-managed Kubernetes, and since then my Ansible skills got a little rusty, but I’m confident I could pick it up again fast, should the need arise.
See also: DevOps.
AWS
See DevOps.
Golang
I’ve been actively wanting to learn Go since 2022. I’ve had the opportunity of writing a few Kubernetes operators with it, and I’m looking for ways to use it more on my work.