Who we are
Dataverse Labs is a European deep tech company. We specialize in spatial data science and spatial data engineering. If data has any dimensions, we work with it – the dimension could be spatial, temporal, or abstract.
Our typical use cases are:
- 🛰️ Remote sensing and spatial data processing
- 🌐 Spatial phenomena modeling
- ⏱️ Time series forecasting and anomaly detection
- 🔮 Sequential events predictions and recommendations
- 💹 Price and route optimization algorithms
- ✈️ MLOps and backend development
- 🪷 Setting custom Machine Learning and AI models in production in cloud environments: AWS, Google Cloud, or on-premise machines
- 💟 Open Source and Open Science
Services
The company is specialized in spatial data science and data engineering. We have created and written multiple machine learning models, data pipelines, and applications for various range of customers and use cases: e-commerce, retail, transportation and logistics, public health, airport services, leisure and tourism.
Our scientists and engineers have developed geostatistical Open Source software. We closely cooperate with organisations responsible for the high-quality scientific Python packaging ecosystem (pyOpenSci), one of the biggest programming school in Poland (CODE:ME), and local NGOs (Hackerspace Trójmiasto).
Statistical, ML and AI Models in Production
We prepare your statistical models to work as a part of a bigger service on the Amazon Web Service, Google Cloud Platform, or your machines.
We set into production the most common models and custom solutions not implemented in any popular AI package.
What does it mean to set a model in production?
- Train it automatically,
- Monitor its performance,
- Serve it as a service.
Modeling and ETL pipelines
We take responsibility for data processing pipelines and complex AI models. Our researchers have developed multiple custom models for e-commerce, logistics, retail, and public health applications.
What are the topics that we are strong in?
- Prediction of the future, modeling and fast detection of anomalies,
- Analysis of the neighborhood, relations between people and their close environment,
- Optimization systems, including dynamic pricing.
Backend Development
Setting models in production cannot be done without backend expertise.
We build simple backend services around your model using AWS serverless functions or your machines.
If your model is a service, we can quickly transform it into a mobile/web application.
Technologies
Core
- Python
- JavaScript
- Matlab/Scilab/Julia
- Git
- Docker
- Databases: PostgreSQL & PostGIS, MySQL, SQLite, MongoDB
- MLFlow, Apache Airflow
- Systems: Linux, MacOS
- Backend development
- QGIS
- ML pipelines
- Systems architecture
Cloud
- Google Cloud: BigQuery, Dataform, Vertex AI, Storage, Functions…
- Amazon Web Services: DynamoDB, Lambda, API Gateway, SageMaker, CloudFormation…
- VPS and custom pipelines
Science
- Geospatial: GIS, remote sensing, network analysis, algorithms
- Machine Learning: unsupervised, supervised, computer vision, time series
- scikit-learn, statsmodels, Keras, TensorFlow, PyTorch, OpenCV, BigQuery ML, pyMC3…
- Genetic algorithms
- Optimization and simulation
- A/B testing
- Publishing Python packages
Portfolio
Statistical Models in Production & Backend Development
- 2024: Custom recommender system in Google Cloud, Dataform, Vertex AI
- 2024: Web application with spatial & temporal forecasts
- 2023: Weather data gathering systems,
- 2023: Discord bot backend on AWS for sentiment analysis
- 2022: Airport Ground Handling app backend on AWS
- 2022: Custom recommender system integration on AWS
- 2021: Route planning app for the waste disposal company
Data Science and Machine Learning
- 2024: Infection Risk Forecasting
- 2023: Sentiment analysis from Discord messages
- 2022-2023: Dynamic price optimization system for hotels
- 2022-2023: Demand forecasting & anomaly detection
- 2021-2022: Custom recommender systems for e-commerce
- 2021: Customer segmentation based on customer behavior and spatial indices
Open Source & Open Science
- pyinterpolate: Pyinterpolate is the Python library for spatial statistics : JOSS publication
- Publication (remote sensing, spatial statistics, public health, vector-borne diseases): High-resolution vector-borne disease infection risk mapping with Area-to-Point Poisson Kriging and Species Distribution Modeling
- Realized projects financed from the European Funds and European Space Agency
Contact info
Szymon Moliński Dataverse Labs
Leszczynskiego 4 room 29
50-078 Wroclaw, Poland
simon@dataverselabs.com
EU-VAT: PL1990086721
Github: https://github.com/DataverseLabs
