You’re in good company

Several American and Norwegian leading companies trust us to help them grow. See how others have realized their goals by partnering with Sannsyn.


Data Analytics, Big Data Pipelines, Statistical Testing: Partnering with 4PatientCare we worked to develop statistical tests to evaluate the efficacy around targeted marketing campaigns, which resulted in a monthly attributable revenue of $90,000 for their largest client. Sannsyn was responsible for all statistical experiment design, infrastructure development, and algorithms in order to quantify hypotheses surrounding medical patient retention. We created HIPAA compliant ETL pipelines and a novel Bayesian A/B testing model in order to robustly quantify uncertainty estimates around patient conversion rates.


Deep Learning, Natural Language Processing, Data Mining and Augmentation: Sannsyn worked with Avochato to develop and deploy a scalable, real time non-binary sentiment classifier for use in both their mobile and web apps. Utilizing the latest developments in deep learning, we created API endpoints to a custom model which ingested incoming messages/conversations (text and emoji support) and returned a sentiment score to quantify how positive or negative input conversation was. A novel methodology based on semantic -similarity was developed in order assist and automate the labelling of training data. The resulting product was a set of scalable API endpoints deployed in a docker environment that would preprocess and serve data to the classifier.


AI Models, Signal Processing, Deep Learning: Pzizz is a sleep app that is widely used globally by users that include J.K Rowling and Roy Hibbert. We partnered with Pzizz for an R&D project around utilizing bleeding edge deep learning methodologies (GAN’s and Autoencoders) for applications in generative signal processing and time series denoising. We developed novel data augmentation methodologies and robust GAN pipelines, which included both traditional signal processing techniques and custom methodologies, for signal cleaning and generation.