Leveraging managed services

Samuel Couture Brochu
XpertSea Engineering
3 min readFeb 9, 2018

Ideas are a dime a dozen. Bringing them to reality is a whole other story. When I joined XpertSea, the company had a ton of ideas, but no in-house technological expertise. The goal was to bring reliable, actionable data to the aquaculture industry. But where to start?

We hired as much as we could afford, building out the core of our engineering force. We worked hard and learned a lot together. In the process, we:

  • Launched the XperCount, a portable, connected measurement device capable of running embedded computer vision;
  • Delivered a state-of-the-art Web platform to harvest and analyze the data;
  • Built the foundation of the first and largest aquaculture data source and data collection pipeline;
  • Trained advanced artificial intelligence models to detect a variety of different aquatic organisms using deep learning;
  • … and just built a lot of stuff.

This was done in less than 2 years and with a core team of 5 developers. We are not the mythical 10x developers. Nor do we work 60 hours per week. We make mistakes. In fact, we make a lot of mistakes. Yet we managed to build game-changing products that found buyers in 48 countries in such a short time frame. Sometimes, when I look back, I wonder how we did it. And, of course, when I look forward, I see how much remains to be accomplished.

The Cloud

There a lot of options when it comes to hosting your data and selecting your computational power. The Infrastructure-as-a-Service sphere is quite crowded. All are rather low maintenance and hassle free. Perfect for any small (and busy) team.

We chose Amazon Web Services. Why not Microsoft Azure or Google Cloud Platform? AWS offers a more competitive combination of services and pricing. We also strongly feel they lead the pack when it comes to IaaS providers.

For us, there are numerous advantages to using a cloud service like AWS (or any other):

  • Simplify our daily lives. Deploying a new PostgreSQL database or running a Docker container is so easy.
  • On-demand resources with near infinite scaling. The research team can start 300 CPUs and a few GPUs to train a model. When the job is done, all the instances shutdown and we stop being billed. Handy, right?
  • Easy replication and geo-redundancy. Instances or storage can be easily redeployed and duplicated in other locations. Some manage services offer automatic traffic redirection and replication.
  • Skip ownership problems. Remote, third-party hardware is never obsolete. You don’t have to buy things in case you need them. You pay-as-you-go. Costs are market driven. No storing backups at the bank. No heating warnings at 2 AM. No internet or electricity outage.
  • Flexible offering. Whether it’s a container service, serverless code execution, databases, DNS, CDN, VPC, API Gateway, they have it all.

But the most important one:

  • Focus on business logic. We are not infrastructure gurus. However, we’re pretty good at other stuff. So we‘d rather spend more time on our core competencies, and leverage third-party services which allow us to maximize efficiency.

Serverless

Our git repositories run on Bitbucket cloud and our continuous integration and deployment server is Semaphore, a cloud service. We use Chartio, a cloud-based business intelligence software. Docker Hub for our containers, Google Suite for our office needs, Asana for our product development, Zendesk for our customer support… We often joke about our server room being serverless. It’s all good, it was re-purposed as a brewery and whisky cabinet.

The Dream

“In the future, all the code you ever write will be business logic.” — Werner Vogels, AWS re:Invent 2017.

You might say this is a distant future, but I think it’s closer than we imagine. As for now, we’ll take what’s available. If there is one takeaway from our experiences, it should be to not reinvent the wheel, leverage managed services and use what the cloud has to offer.

Startups constantly evolve. We came a long way as a company, as a team, and our dreams grew with us. To keep everyone fed, aquaculture production value needs to double to reach 400B by 2030. We aim to spearhead that Blue Revolution, where high-tech, data-driven aquaculture helps feed the planet. We can’t do it alone.

-

Looking to join an awesome team building cutting-edge, world positive tech? Apply on our career page: https://www.xpertsea.com/careers.

--

--

Samuel Couture Brochu
XpertSea Engineering

CTO at Xpertsea. I spend most of my days working with a bunch of passionate engineers. My interests are whisky, blockchain, robotics, AI & machine learning.