About neXDos
neXDos is a Germany-based asset management company primarily focusing on listed equities. With its team of data professionals and industry experts, neXDos utilizes extensive quantitative research and data analysis to help manage clients’ investments efficiently. The company has recently launched its first public investment fund, marking a significant milestone in its journey.
Zero Data Science Platform Downtime – neXDos Says It’s Real With Datalore
JupyterHub Hosting Hurdles
neXDos heavily relies on diverse data sources – from stock market APIs to in-house PostgreSQL data – to inform their bespoke investment strategies. Previously, neXDos prototyped, backtested, and future-proofed their trading algorithms by hosting JupyterLab and JupyterHub. But this approach was problematic due to complexity in many areas, including managing the server environment, carrying out frequent updates, and handling the associated plugins within JupyterLab.
“Hosting JupyterHub on a Kubernetes cluster was a hard task, primarily due to the complexities of managing regular upgrades and installing and maintaining numerous plugins.”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
Further complications arose due to security requirements and Germany’s stringent data protection regulations, which necessitated hosting neXDos’ operations on bare metal servers. To maintain compliance, management of user permissions and access rights was critical. This entailed developing custom LDAP authentication scripts – a process that posed complex security risks, especially in granting interns temporary access to certain data.
“Managing access rights was quite a challenge, especially when it came to temporary permissions for interns. Granting access and later rescinding this access often presented security concerns.”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
In the face of these challenges, neXDos required a solution that would fulfill their technical requirements:
- Compatibility with bare metal servers
- Compliance with German data protection requirements
- Simplified user access control, along with easy team collaboration
Enhanced Workflow with Datalore: Streamlined, Collaborative, and Secure
neXDos turned to Datalore as it catered to their needs, offering an on-premises installation, built-in SSO authentication, and streamlined collaboration and permission management within Jupyter notebooks.
“Datalore’s streamlined workflow allowed us to go from strategy prototyping to testing and deployment with ease.”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
Adopting Datalore as a data science platform gave neXDos an efficiency boost in the following steps of their workflow:
- For each new project, the neXDos team would create a new team workspace and grant view and edit permissions to certain team members. The team lead would pre-configure the necessary workspace-wide SQL database connections and environment configurations using init.sh scripts.
- Quantitative analysts would then create a new Datalore notebook to gather all necessary data from stock market APIs and the corporate PostgreSQL database using Python and SQL and prototype a bespoke investment strategy.
- Analysts would then backtest results and future-proof the strategy by using Datalore’s Scheduled runs feature to run a daily analysis.
- After the strategy was proven to be successful out of sample, a Python Engineer would then turn the Jupyter notebook into a Python script and deploy it to perform sell and buy advice orders.
“Datalore’s coding assistance feature greatly enhanced our productivity. It helped us in quick prototyping and made coding in Python and SQL much easier.”
— Henry Eitel, Software Engineer at neXDos
Results
before
8
hours of downtime per month
with open-source JupyterHub and JupyterLab hosted in a Kubernetes cluster
after
0
hours of downtime per month
with Datalore Enterprise hosted in a Kubernetes cluster
“With Datalore, we discovered a reliable platform that provided us with the right balance of collaboration, efficiency, and data security”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
Implementing Datalore as their primary data science platform led to improvements in neXDos’ performance and productivity. The process of prototyping, backtesting, and deploying trading strategies became more efficient.
“Datalore’s collaborative workflows and integrated environment, permission, and data management significantly accelerated our move from prototyping investment strategies to testing and actual deployment.”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
neXDos also found Datalore’s reliability to be beneficial. With zero incidents of service downtime, neXDos’ operations went uninterrupted, boosting productivity.
“Datalore’s uninterrupted service and easy upgrades give me peace of mind and one less tool in our stack to think about.”
— Henry Eitel, Software Engineer at neXDos
Another positive of Datalore is the way its robust permission management features allowed neXDos to incorporate interns into specific projects.
“Datalore’s robust permission management enabled us to invite interns into our projects while adhering to our strict data security obligations.”
— Dr. Steffen Möllenhoff, Managing Partner at neXDos
Similar Customer Stories
Drama & Company
Youngrae Lee, Big Data Center Team Lead at Drama & Company
Before adopting Datalore, Drama & Company used a standalone Jupyter server for its data research needs. However, this setup often faced performance hiccups. Transferring the research results to a data pipeline, distributing the code written by researchers, and other similar tasks proved to be tough and time-consuming. Drama & Company realized the need for a more robust and scalable data science platform.
Constellation Software
Nauman Hafiz, CTO at Constellation
To scale data science operations and provide valuable insights to both clients and internal stakeholders, Constellation needed a way to streamline collaboration and enhance agility. Issues like slow report generation in traditional business intelligence tools (Power BI and Looker, specifically) and disjointed workflows limited their ability to produce timely, customized insights.
LINE Corporation
Seongduk Cheon, a Senior Manager at LINE Corporation
During the evaluation process, we found that Datalore’s UX was familiar to our developers, and the report-sharing functionality was easy to use. Thanks to the collaboration of our engineering team and the Datalore development team, we managed to satisfy our workflow and data governance requirements.