About the hedge fund
Founded in the 1980s the New York-based investment management firm has become one of the world's largest hedge funds with billions in assets under management. Their investment philosophy revolves around quantitative trading and data analysis, which has led to strong returns for clients. The firm has a global presence with offices around the world.
Infrastructure and tooling setup
The fund had migrated to the cloud several years ago, and the operational setup was very mature. While the firm had many resource types, the main applications were hosted on AWS with EKS. They primarily use GitHub Actions for CI and JFrog Artifactory as a Container Registry. They operate a wide range of databases from MongoDB to Postgres, used Redis, Elastic and MariaDB running in cluster, and Route53 to manage DNS. Before adopting Humanitec the fund managed its applications as configuration as code, with Helm and Terraform for Infrastructure as Code (IaC).
Key challenges
Hedge funds are constantly seeking to maximize profits while minimizing risk. One way they do this is by leveraging quantitative trading models. These models use algorithms to analyze vast amounts of data and generate predictions about the market, which can be used to make trades.
In the fast-paced world of finance, time is of the essence. The ability to quickly deploy trading models is critical for hedge funds to capitalize on market opportunities, minimize risk exposure, and remain competitive. The fund was presented with the following key challenges:Â
- Developers were often overwhelmed by the complexity of the underlying setup. This kept them away from writing and testing new models.Â
- Ops bottlenecks caused by key person dependencies slowed down overall delivery. Waiting times for databases, environments, and other resources blocked development.
- Varying configuration formats and missing guardrails made it hard to drive standardization resulting in maintenance overhead.Â
- Quantitative traders weren’t fast enough to deploy their models, the lifeblood of the business.
"For us every hour counts. A model that’s not running can cost us millions."
SVP Operations
Platform architecture
The fund designed its platform following our AWS reference architecture. Using the Humanitec Terraform Provider, the Orchestrator configuration was done as code and as an interface, the firm chose Score to ensure code as the single source of truth. The platform team configured the application baseline configs and built reusable infrastructure templates using Humanitec’s open-source drivers. Developers described their workloads and dependencies using Score and operated/deployed their models staying in their git-push workflow.Â
Key improvements
Introducing Humanitec’s Internal Developer Platform, the hedge fund’s operations teams were able to define clear golden paths to enable developer self-service. The procedure for the developers to get their model in production was reduced to a single file and minutes of work. Developers autonomously self-serve the tech and tools they need in fully provisioned, dynamic environments.
- Zero developer dependency on Ops
- 97% resultant drop in Ops tickets and requestsÂ
- 4x model deployment increased
"The speed at which we are deploying trading models today would have frankly not been possible without Humanitec. This is a huge competitive advantage.”
SVP Operations
Timeline and evaluation
- POC: 6 weeks
- Evaluated against a self-built setup estimated to take 12-24 months.Â
- Total migration: 18 weeks
- Onboarding per new developer: 30 minutes
