TECH STACK
This is how we work everyday.
Code quality
We know that we cannot succeed without high-quality code. We spend a significant amount of time crafting components and services that are efficient, clean, readable and reusable as much as possible.
Tool choice
We love exploring new technologies and we're free to pick the best tool for each task — but freedom isn't anarchy. Some problems deserve a battle-tested, boring solution; others are worth a bet on something new. Knowing which is which is part of the job.
Testing
No user-facing feature goes to production if it's not tested.
We think automated testing is the best way to ensure quality and maintainability of the code, and to make our developers, stakeholders and users happy :)
Security
Many clients, among which are banks, insurance companies and big corporations, trust us with their data, so we take security very seriously. We adopt all the best practices to protect our data, our infrastructure and our assets.
OUR TOOLS
Technically speaking, we mainly use languages like Python, Golang and Scala for the back-end, and TypeScript and React for the front-end.
Our infrastructure runs on AWS and is managed and tested as code with Ansible and Terraform. We use GitLab to store all our code and run our CI/CD pipelines. We rely on Prometheus, Grafana, Logstash, Kibana and Sentry to monitor our applications, and we deploy most of our components on our Kubernetes cluster.
We store our data in PostgreSQL databases, mirror it across several Elasticsearch clusters and cache it on Redis instances. Data is processed through Kafka queues, with Celery as our messaging system. We run Spark and MapReduce jobs — increasingly on Databricks — to clean, match and link data from several sources and providers.
Django and Next.js are our main web frameworks. REST APIs and gRPC endpoints are our bread and butter.
Our data science team relies heavily on Jupyter to experiment with data and algorithms, uses tools like scikit-learn and TensorFlow to build machine learning models, crafts AI agents with LangGraph, and works with Langfuse, LiteLLM and vLLM.
