We worked with a public medical device company to build a machine learning platform that enabled data scientists to put models into production—the forecasts of which are now leveraged by downstream applications.
A public pharmaceutical company wanted to better understand their prescribers, so they could have more productive conversations and create better solutions. We helped them use the data they already had to set everyone up for success.
In 2020, we partnered with a wearable device maker to use wearable medical data to predict illness—even before symptoms showed up.
We partnered with a payer data platform company to aggregate and process disparate health insurer data, enabling payers to run analytics on all of their data in one place.
We worked with a real estate investment company to discover and build an internal product that puts all their data layers in one app with a snappy interface, allowing them to make better capital outlay decisions.
We take machine learning models from experiments in a Jupyter notebook to production in the cloud. We follow a structured MLOps process for the entire modeling lifecycle, including Dockerized ML development, parallelized backtesting, ML API patterns, model explainability, model performance monitoring, and infrastructure as code modules for rapid deployment.
We develop models to solve hard problems—from unsupervised anomaly detection in multi-variate time series to dynamic system identification using deep learning. We take a heterodox approach to data science: we start from first principles about the mathematical formulation of the problem and then experiment with the relevant methods—from modern hierarchical Bayesian methods, to gaussian processes, to deep learning, and sometimes to more classical techniques like Kalman filters. As with all good science, we start simple, experiment a lot, and iterate our way to the best possible solution.
We engineer modern data infrastructure, including purpose-built enterprise data platforms, complex data pipelines, batch and streaming data, sensitive data handling, serverless technologies, and more. In data science and machine learning, more and better data always beats better algorithms.
We bring a unique perspective and expertise to our work. In this collection of essays, papers, and conference talks, we lay out our core philosophies and approaches. Explore the fundamental DNA of what makes us Manifold, learn about some of the tools we use every day, and understand how we build successful data products. Some of our most popular pieces include: