Your partner in analytics

About the company

At SPRYFOX we believe that there is potential in any kind of data. Most companies gather huge amounts of data without benefitting from it. We set our mission to help those companies understand and utilize their data in the most optimal way. At SPRYFOX we have strong expertise in data analytics and IoT products. SPRYFOX builds solutions with a differentiating factor for you, analytics which amaze your customers and tools that improve your day-to-day processes.

"If you can’t explain it simply, you don’t understand it well enough."

SPRYFOX explains it simple.

If you want your data explained simple...
Contact us

Our services

End-to-end solution development

Our specialty is that we can cover the whole expertise needed to build end-to-end systems, covering all aspects from sensor evaluation and selection, sensor and data integration and robustification, lower level analytics, artificial intelligence and other higher level analytics up to visualization and dashboard creation. This allows for minimal friction to run a project having all technical components from end-to-end targeted to achieve the respective business goal. At SPRYFOX we are happy to develop the whole or also parts of the end-to-end solution.

Analytics module development

Often, an end-to-end system is already existing on customer side, but additional functionality is required to e.g. enable more data-driven features. These are typically specialized analytics modules such as an anomaly detection or predictive analytics that are running in the existing customer ecosystem. At SPRYFOX we are agnostic with respect to the existing infrastructure and provide targeted modules, working in very close collaboration with the customer.

Proof of concept

Especially when the business value and/or the required data to achieve this business value are not fully understand yet, it is wise to first invest into a proof-of-concept. Our proof-of-concepts are driven by answering three questions: a) quantification of the business value for the system in mind, b) determination of the required data sources and needed data quality, c) determination of the suitable analytics to reach the business goals. With these questions in mind at SPRYFOX we are able to conduct very targeted proof-of-concepts that within reasonable time provide the necessary basis to decide about further steps.

jw
cd
sm
em
sb

The Team

Our core team

We have been leading and executing industrial data science and IoT projects since 2010 in various industries.

Data Science

Machine Learning, Deep Learning, Big Data Analytics, Data Mining, Text analysis/NLP, Data & Feature Engineering, Statistical Analysis, Image & Video Analytics, Reporting, Data & Model Quality Assurance, BI Dashboards, Explainable AI, Fair and Unbiased Machine Learning, Data Visualization

Programming, Data Analytics Tools & Cloud

Python, Java, WebTechnologies (JavaScript), Talend, Dataiku, Tableau, Power BI, Amazon Web Services, Google Cloud Platform, Microsoft Azure, Snowflake, Edge Development, MATLAB

Founders

Johannes Wowra

Founder

Christian Debes

FOUNDER

Management

As your partner in analytics we work together with you in defining and building the right components and systems to achieve your goals. With a strong business focus in mind we are typically working in an agile way with fast iterations and a customer-centric project management. Depending on the business need we provide different models on how to work together in a flexible way: From extending your existing data science teams up to responsibility for full end-to-end solutions. 

Application Requirements

At SPRYFOX we are able to create powerful analytics on various production environments. Development on edge devices, mobile devices, on-premise systems and cloud environments are our daily business. We can handle a plethora of data sources, including but not limited to: sensorial, physiological, machine logs, legacy systems, images, audio, etc.

At SPRYFOX we are able to create powerful analytics on various production environments. Development on edge devices, mobile devices, on-premise systems and cloud environments are our daily business. We can handle a plethora of data sources, including but not limited to: sensorial, physiological, machine logs, legacy systems, images, audio, etc.

Analytics

Classification

Classification is all about learning models for specific situations/behavior that are recurring in your application. Instead of using rules to detect them we train a machine that learns the specific constellations under which these situations/behaviors occur in reality. Typical examples are images, sound events, machine states or human actions. Our classifiers are designed to explain the reasoning to a non-technical user and improve over time.​

Clustering

Clustering is a more exploratory technique that allows identifying common structures in data to then exploit them in further applications. For example, it allows to tremendously reduce the need for labeled data in classification.

Data Mining

Often we are facing the situation in which a lot of data from various sources is available, but we are unclear about what value it can bring or which use cases it could realize. In these situations, data mining as an explorative technique is perfect to detect hidden patterns and relations between those data sources. These patterns are then used as data-driven use case generators whose validity can be checked against relevant business parameters. ​

Anomaly Detection

Anomaly detection models the normal behavior/situation and uses deviations from that normal model to reliably detects anomalies. Our anomaly detection engine is designed to put explainability (why did this anomaly pop up?), confidence (how sure are we that this is an anomaly?) and continuous improvement (using user feedback on relevance and severity of said anomaly to train better and better models over time) in the focus.​

Predictive Analytics

In many practical situations we are interested in learning models of the past and present to predict likely outcomes for the future. Our predictive engines allow joint handling of continuous and event data and takes different temporal dimensions into account. This allows building several models, e.g. for the near or mid-term future. Typical applications are failure prediction of machinery and customer conversion.​

Deep Learning

Some of the biggest breakthroughs in artificial intelligence in the recent years were achieved via deep learning. Learning deep neural networks allowed for super-human performance in several areas including image and speech recognition. We use deep learning methods in many different fields while being well aware of its limitations in some practical fields where labeled data is scarce and explainability is more important than performance.​

Classification

Classification is all about learning models for specific situations/behavior that are recurring in your application. Instead of using rules to detect them we train a machine that learns the specific constellations under which these situations/behaviors occur in reality. Typical examples are images, sound events, machine states or human actions. Our classifiers are designed to explain the reasoning to a non-technical user and improve over time.

Clustering

Clustering is a more exploratory technique that allows identifying common structures in data to then exploit them in further applications. For example, it allows to tremendously reduce the need for labeled data in classification.

Data Mining

Often we are facing the situation in which a lot of data from various sources is available, but we are unclear about what value it can bring or which use cases it could realize. In these situations, data mining as an explorative technique is perfect to detect hidden patterns and relations between those data sources. These patterns are then used as data-driven use case generators whose validity can be checked against relevant business parameters.

Anomaly Detection

Anomaly detection models the normal behavior/situation and uses deviations from that normal model to reliably detects anomalies. Our anomaly detection engine is designed to put explainability (why did this anomaly pop up?), confidence (how sure are we that this is an anomaly?) and continuous improvement (using user feedback on relevance and severity of said anomaly to train better and better models over time) in the focus.

Predictive Analytics

In many practical situations we are interested in learning models of the past and present to predict likely outcomes for the future. Our predictive engines allow joint handling of continuous and event data and takes different temporal dimensions into account. This allows building several models, e.g. for the near or mid-term future. Typical applications are failure prediction of machinery and customer conversion.

Deep Learning

Some of the biggest breakthroughs in artificial intelligence in the recent years were achieved via deep learning. Learning deep neural networks allowed for super-human performance in several areas including image and speech recognition. We use deep learning methods in many different fields while being well aware of its limitations in some practical fields where labeled data is scarce and explainability is more important than performance.

Contact us

Want to understand and discover the potential of your data? Contact us and we will help you discover the diamonds in your data heap.

Contact us

If you're looking for a partner to realize your analytics vision.