data science, automated systems and artificial intelligence.
We are a young company based in Stuttgart
that accompanies you on your way entering the field of artificial intelligence and turn your
ideas into reality.
What We Do
Consulting
Provision of targeted information and expertise on the topics artificial intelligence,
data-driven development and technical computations.
Projects
Data analysis, feasibility studies or support for larger AI projects.
We support your company in the implementation of future visions around new development
approaches or intelligent products.
Training Courses
Group training or customized solutions for beginners and advanced users in
the fields of machine learning, data analysis and automation.
A Qnovi Product
codingQ
An on-prem framework for deploying an in-house AI programming assistant.
CodingQ is an AI programming assistant that is based on large open-source large language models and is
operated as a server-side service locally on the customer's IT infrastructure. Sovereignty, data
protection and confidentiality are thus guaranteed and companies retain complete control over the
use of their data.
Qnovi provides support throughout the entire process, from planning and hardware acquisition to an
optional test phase and production deployment. CodingQ generates code, converts it into other
programming languages, supports troubleshooting and can be used to automate routine tasks such as
test generation and documentation.
In the context of ocean-going vessels, the choice of the ship's route as well as the cruising
speed has a major impact on the economic and environmental costs.
Fast digital twins of the ship propulsion system based on neural networks are a basic
requirement, in order to be able to investigate approaches such as reinforcement learning (RL)
for solving such complex optimization problems based on them.
In our joint publication with WinGD, we demonstrated the value co-creation potential of such
digital services.
A second focus was the comparison and linking of simulation and real ship data.
In our blog post, we show how to adapt large open source language models and create your own
German language model. Here we go into the text preparation, the pretraining, the translation
and the final fine-tuning.
With this, we can create a language model that can write in the style of Wolfgang von Goethe's
Faust and beat other larger German commercial language models.
Qnovi GmbH created an AI algorithm for predicting the electricity demand of a commercial
facility as part of a competition.
The forecast for electricity generation with the company's photovoltaic system considers the
influence of the weather, while the consumption forecast takes into account operational
processes and human influence factors.
The model forecasts can be used to determine how much residual energy must be procured on the
EPEX day ahead or spot market.
Multiple years of experience in the field of supervisied learning and reinforcement
learning for prediction and imporvement of complex technical processes.
Experienced in Tensorflow, OpenAI Gym and distributed learning.
Dr. Dominik Rether
CEO
CEO of Qnovi GmbH with long-term experience in software development with Fortran.
Insights into machine learning and AI
and routine in coordinating research and industry projects.
Dr. Valerian Cimniak
AI expert
Several years of experience in the acquisition and processing of simulation and measurement
data for time series prediction and anomaly detection.
Experienced in using Tensorflow as well as building automated computational analysis tools.
Cookies
Did you miss the annoying cookie question box?
Our visitors should be able to browse freely without being captured by advertising, social media or
analytics cookies.
If you write us via the contact form, a technically necessary cookie is set.
