Prediction and forecasting
Models that forecast demand, prices, risk or failures from your historical data.
Service · Prague & remote
Prediction, classification, computer vision and data analysis with TensorFlow, Keras and PyTorch, for businesses in Prague and across the EU. My bachelor's thesis at CTU predicted urban heat maps from satellite images with convolutional neural networks, and at ECOTEN I built machine-learning models for thermal data prediction.
Models that forecast demand, prices, risk or failures from your historical data.
Recognising objects, defects or text in images, scans and video.
Clean, automated data flows and dashboards that people actually use.
A small, measurable experiment on your data before any larger investment.
Models wrapped in APIs, monitored and retrained — not left in a notebook.
Results explained in business terms, including where the model is unsure.
Real projects from my portfolio that this service builds on.
A 30-minute call or meeting about your goals, customers and budget.
A clear scope, timeline and price — before any work starts.
You see and click through the solution early and shape it with feedback.
Built in short iterations with regular previews, tested on real devices.
Deployment, monitoring and help after launch — plus maintenance if you want it.
It depends on the task. Some problems work with a few thousand records, others need more — or can start from a pre-trained model. A proof of concept tells us quickly.
Classic machine learning is best for predicting numbers and categories from your data; generative AI for text, documents and conversation. Many projects combine both.
Yes. Models can be deployed in your cloud, on your own servers or on devices, depending on data and latency requirements.
Tell me what you need — a free 30-minute consultation, no obligation. I reply personally.