Discover ways to apply machine learning to your cause

Mission

Simple-AI opens a gate for users to readily apply state-of-the-art machine learning technologies to their cause, in a cost-efficient manner.

Algorithmic solutions at Simple-AI are modularized and adaptable to a wide spectrum of data-driven real-world applications. In particular, we design highly flexible AI algorithms for perception, prediction, planning, reasoning and control.  Simple-AI platform is open source.

Why shall we care and use Simple-AI services?

The life cycle (design, development, deployment, expiration, replacement) of an AI algorithm is tremendously shorter than conventional programs. This phenomenon has clear origins and far-reaching implications for our society. Their short life-cycle is the result of the following:

Financial Resources

A large concentration of budget from the public and private sectors goes into AI research and development. This has significantly accelerated progress in the field.

Compute

Technological advancements on high performance and parallel computing infrastructures

Accessibility of Code

Our field is open source. This availability of code and resources to everyone results in the evolution of algorithms faster than ever.

Scientists

The brilliant pool of talents worldwide connected through digital mediums has produced innovative and diverse algorithms that are getting improved on a daily basis.

The Opportunity

This infrastructure provides industries of all sectors with a unique and exciting opportunity to improve (from a cost, energy, and time perspective) their data collection, data processing, quality control, predictive analysis, and autonomy pipelines.

The Challenge

It is now much harder to identify what AI algorithm is most suitable for a given use-case, simply because the space of possible choices is vast, complex, and dynamic. Nonetheless, new and better machine learning algorithms are getting developed on a daily basis; how can we keep up with the pace of these advancements?

A brute-force approach that is common nowadays is to try out models from open-source repositories and perform exhaustive training and tuning on in-house or cloud-based compute servers to find a model that best fits data. This method is very cost-inefficient and slow and can easily become outdated.

For instance, a small-sized insurance company that seeks a deep learning-based document summarization algorithm to analyze text data, with a brute-force approach would have a total first-year cost of around €800k – €4.6m for taking a performant pre-trained language model, tune it on their own data and actively retrain it with new data. This is where Simple-AI comes in.

The Simple-AI Process

We use our years of expertise in developing advanced machine learning models and our efficient machine learning package to save costs and time for our clients. For example, the insurance company above can save up to 60% in costs if they become Simple-AI clients.

  1. Professional data evaluation and curation via state-of-the-art self-supervised and augmentation strategies

  2. Algorithm selection from the most computationally efficient deep learning models to date

  3. Efficient tuning via our native hyperparameter tuning engine for advanced models

  4. Extensive testing and model validation at inference, via recurrent online and offline inference methods

  5. Delivery of model candidates and ensembles.

Team

Philipp Neubauer | CEO & Founder
Ramin Hasani | Founder
Sophie Grünbacher | CTO
Mathias Lechner | Founder


Contact us

For project inquiries and careers at Simple-AI please get in touch with us:

contact@simple-ai.com