Engineering at Amplitude
We’re building the future of product.
We have a growth mindset.
My manager and tech lead solicit my opinions and feedback regarding my career trajectory. They both work incredibly hard to give me the opportunity to learn.
We aim to raise the bar.
Since we have reached an important milestone of 200B events per month, collaboration is more crucial now than ever so we can scale up our cutting-edge applications. I am very excited to be part of this challenging project with the team.
We ship fast.
Shipping fast at Amplitude means getting feedback from customers early in the product development cycle. Better to test something early than spend months on a problem and then realize it was the wrong problem to begin with.
We aim to win.
The level of ownership people across the team have for their work is really special. You can really sense the effort and care that goes into what they do, which is incredibly motivating.
We are customer focused.
Our team has a tremendous amount of empathy for our customers and a desire to succeed as a team. Our solutions are solving real problems our customers are facing.
We take risks.
We’re big fans of running experiments. Whether it’s a new feature in the product or an interesting new technology that helps us build it—experiments help us find the right bets that could really change the trajectory of the company.
Getting customer data to Amplitude
Ingesting 150,000 events per second
We collect and process over 200 billion data points every month, peaking at over 150,000 events/second. We’ve built a horizontally scalable, highly available system that can smoothly scale up and down with incoming traffic. We make this happen by leveraging great open-source technologies like Kafka, PostgreSQL, Redis, Cassandra, and ElasticSearch.
Querying tens of billions of events
At the core of Amplitude is a flexible and scalable query system that enables customers to ask questions about their data. Written in Java and backed by Amazon S3, our proprietary distributed column store Nova, is designed for interactive behavioral analysis. It regularly serves queries running on tens of billions of events, with a median query time of <1 second. We’re constantly innovating on its functionality and efficiency.
Presenting the data
The frontend is our centerpiece–it’s how we turn a traditionally complex and intimidating experience into an empowering one. We build experiences focused on exploration, understanding, collaboration, and sharing—recognizing that an important insight from your data is only as valuable as your ability to share it with your organization. We work with our talented design and product teams to build fast, reliable, and intuitive user interfaces. Our web application is built on a modern stack of React, Redux, Highcharts, D3, and node, and speaks to a Python API server.
The infrastructure underneath it all
In the analytics space, operating securely at scale is non-negotiable. We run over 1000 instances in Amazon’s Virtual Private Cloud with a lean engineering team, and we regularly deploy over 50 microservices that power the product. Our DevOps efforts rely on a suite of powerful automation and monitoring tools: Docker, SaltStack, Terraform, DataDog, PagerDuty, and Jenkins.