Our Client had:

Traditional approaches to anomaly detection in grid networks are operator-based in a grid operating center. Operators use collected near real-time metrics from the grid to assess and anticipate anomalies. The operator must decide, based on experience, which mitigating measure to use and lacks the ability to simulate scenarios in real-time.

Our Solution:

Hivemind implemented a near real-time anomaly solution based on Apache Kafka and Apache Spark, processing 3.6 million data points per second to determine anomalies in the current and voltage data points and fingerprint anomalous data sets for simulation of downstream predictive systems.

Our Client Needed:

Implement an anomaly detection service with fingerprinting in energy grid networks. 

Our Client Needed:

Due to the success of their CBIS platform, management wanted to see new reporting and APIs. Therefore, the platform must now be always available and reliable to serve these new requests. Anomalies and problems from the source systems must be identified quickly, and data must be up-to-date at all times.

Our Solution:

Hivemind supported PAYONE in the design, operation, and implementation of the CBIS platform. Its goal was to unify the different source systems gathered through various mergers into a uniform reporting tool with a central data model. To achieve this, we introduced architectures for near real-time reporting and streaming, as well as cloud services. By using elastic and scalable services in public clouds, we have also implemented cost reduction solutions. The PAYONE team has also been trained on the new architecture and technologies in order to independently maintain the platform.

Our Client Needed:

The main obstacle for EV adoption is range anxiety and the availability of charge points. The charge point market is heterogeneous and consists of over 700 providers in Germany alone. The quality of service and availability greatly differ, thus hampering the experience for EV drivers. Elvah introduced a scoring platform that gives users the ability to determine the quality and availability of charge points from their app.

Our Solution:

Hivemind implemented a scoring solution for Elvah that uses real-time and historic session data for all charge points in Central Europe. Using machine learning-based predictions, it allows users to determine the availability and quality of charge points at any given time. Hivemind implemented the training models, an MLOps delivery pipeline, and a REST endpoint for serving near real-time predictions using Apache Spark, Spark ML, and microservices running in Kubernetes on AWS.

Our Client Needed:
Hays wanted to create a job board web app to link their clients with potential applicants. The app also had to integrate with their legacy systems.

The web app has two sides:

Talent Applicant Side: Applicants can create profiles containing their skills, view and track job offers, receive client propositions, schedule interviews, and even track time after being hired.

Client Side: Hays' primary clients can create job ads, specify the kind of profiles they're looking for, contact applicants, and schedule interviews.

Our Solution:
We created the entire stack, including the frontend (Flutter), backend (Scala, microservices, Kafka), and the infrastructure (Kubernetes).

We implemented a matching service between applicants and job offers.

From the beginning, we not only developed the solution but also trained their developers. Starting with four developers and growing to over 30, we integrated them with our team throughout the process.

Crucially, we trained them in Test-Driven Development (TDD) and set up pipeline merge rules. This drastically reduced the amount of overtime work they were doing.

Why choose Hivemind?

Benefit from the longstanding experience of our senior development team.

We believe in sharing knowledge to empower your own teams.

We always pursue a best-practice approach, striving to create things as simple and efficient as possible.