Market-surveillance platform: 20+ data sources, 20–80 events per second

Project goal

Client: a US fintech company. We worked on its market-surveillance platform through a partner and do not name either. Years: 2021–2026.

The platform watches trading activity for manipulation and compliance breaches. It ingests data from many venues without pause. Analysts need related alerts grouped into cases they can act on.

Solution

Our engineers worked inside the client's product organisation, directly with product managers in the US:

• New microservices for case aggregation and notifications.
• Proofs of concept and integrations.
• Optimisation of long-running operations in production, and CI/CD.
• Stack: Spring Boot services in Kotlin and Java, Kafka (Confluent), PostgreSQL, Snowflake, Elasticsearch, Redis and Angular.
• Platform: Kubernetes with Helm and ArgoCD, GitLab CI, Airflow, Prometheus, Grafana and Jaeger.

The platform has about 24 services. We built part of them.

Results

• The platform in production: 20+ data sources and 20–80 events per second.
• Production deployments in the EU and the US.
• 99.99 % platform uptime.
• Our engineers worked on it from 2021 to 2026.

Why it matters for you: If you run an event-driven system that a regulator may question, our engineers worked on one in production from 2021 to 2026.