Turku, Southwest Finland, Finland
Data scientist specialized in machine learning and computer vision. My background is from research and development work via AI solutions, which I have worked on both in the public and private sectors. My main job responsibilities during the last years have revolved around analytical problem solving via machine learning and statistical methods, project management, producing intelligent cloud-based software solutions (Azure DevOps + IaC + Kubernetes) for institutions and client companies. My work also involves coordination and teaching of machine learning experts and methods, grant writing, organizing seminars and public presentations. My Ph.D. (minors in physics and mathematics) studies were focused on the development and application of machine learning techniques for open remotely sensed data sets. Current research focuses on projects involving GIS and remote sensing data (peatlands, forestry, LiDAR, CHM, Sentinel, DEM, Hyperspectral data and derivatives). Theoretical studies involve applications of Bayesian optimization in survey sampling and development of new solutions in neural network context. I have a passionate interest in all the sciences and technologies related to artificial intelligence, and I actively increase my frame reference on these subjects like the latest software, algorithm and theoretical solutions. - Top skills - ◼️ Data analysis and machine learning ◼️ Computer vision ◼️ Remote sensing (hyperspectral data, LiDAR, Sentinel, SAR, drones) ◼️ Mathematics and statistical modeling ◼️ Software engineering, cloud-native applications, HPC ◼️ Digital signal and image processing ◼️ Model validation - Latest applications - ◼️ Cloud-native AI-pipeline for enhancing digital data processing ◼️ Automatic classification of plant seedling using robotic arm and computer vision ◼️ Classification of fish species from hyperspectral/RGB imagery. Species distribution and tracking. ◼️ Finland-wide prediction of peatland extent, depth and type with remote sensing data ◼️ Automatic horizon detection for camera calibration in ships ◼️ Situational awareness system combining LiDAR, radar and RGB imagery
I currently work as a senior scientist and coordinator of artificial method services in applied statistics group at Luke. In addition to ML related analysis and software development work, my tasks include e.g. coordination of machine learning method services, project management, arranging machine learning courses and proposal work / coordination of Horizon EU projects. My recent project works include development and project management of Azure-based computer vision automation solutions for natural resource management, full pipeline from data collection, annotation and validation to AI utilization. See my presentation of Luke NatureWatch at: https://www.youtube.com/watch?v=4bbZ0vB4P88&t=60s Several remote sensing based AI projects and computer vision based image classification, such as: automatic detection of fish species from hyperspectral imagery and decision making in robotics. GIS-related projects involve the designing and implementation of automated pipelines of following sort: GIS input data preprocessing --> machine learning based analysis and optimization, feature selection and analysis, nation-wide high-resolution prediction --> automatically produced analysis report. Also, the organization of machine learning courses, such as "Introduction to machine learning" and "Machine learning with Python". Used tools: Python, JavaScript, PostgreSQL, Azure, Azure ML, GitHub, Linux, Docker, CSC, PyCharm, Visual Studio Code, Jira/Confluence, Agile CI/CD, ArcGIS, ArcMap, Azure DevOps, Kubernetes and helm, IaC-Terraform.
I worked as a senior researcher in the Department of Computing in the University of Turku. My research focused on the application and development of machine learning methodologies using drone technology in agriculture and forestry. The research involved the application and R&D of the techniques of data analysis, machine vision, digital image- and signal processing. The main tools utilized in my research were: deep learning, structure from motion, wavelet/Fourier-analysis and local binary pattern. I have passionate interest towards both the applicability and theory of the methods of data analysis. Software engineering is mostly implemented in Python- or Matlab-languages, but I have also experience from Java and C++ frameworks. With respect to the theory of data analysis, I'm interested on the fundamentals of probability theory, measure theory, statistical learning theory and mathematical optimization. Link to my doctoral dissertation: http://urn.fi/URN:ISBN:978-952-12-3710-2 Link to my list of publications: https://research.utu.fi/converis/portal/Person/1097017?auxfun=&lang=en_GB
I worked as a PhD candidate in the Department of Future Technologies in the University of Turku. My research topic focused on the application and development of machine learning methodologies for open multisensor Big Data. Applications of my research included e.g. pattern recognition, regression analysis, image classification from satellite and airborne scanned imagery (COSMO-SkyMed, LiDAR) and route selection for forestry operations based on the predictability of soil bearing capacity. My doctoral dissertation defense was held in june 2018. My responsibilities also included the supervision for Bachelor and Master level theses and lecturing in the courses of Department of Future Technologies (e.g. Applications of Data Analysis course). My working environment was mostly implemented in Matlab or Python environments.
As a project researcher I studied the predictability and classification of forest soil types using public open data (satellite and airborne imagery). My tasks also included preprocessing of the data. Work was conducted mostly on Matlab-environment.
R&D and application of statistical learning AI-systems in autonomous solutions. My work responsibility is focused on the design, implementation and validation of machine learning (object detection/semantic segmentation/regression) applications in Python/C# environments. - Used AI frameworks/technologies -: TensorFlow (CPU/GPU), convolutional neural networks, Scikit-learn, linear regression, k-nearest neighbor, data clustering, cross-validation, matrix algebra. - Used software frameworks -: Python, Jupyter Notebook, C#, .NET Core, Docker, Azure Functions, TDD, NSubstitute, Fluent Assertions, xBehave, xUnit, Jira, JetBrains Rider.
My primary responsibilities consisted from machine vision and artificial intelligence engineering for commercial ships (previous Rolls-Royce ship intelligence unit). The work assignments included the design and software engineering of object detection, sensor fusion and automatic camera calibration via multiple view geometry. The applied sensor technologies included high resolution stereo cameras, RADAR, LiDAR, AIS and INS. The work was implemented in a Python/C++ environments using third-party packages such as TensorFlow object detection, jupyter notebook, opencv and scikit-learn. The software engineering work was conducted utilizing ALM/DevOps frameworks with CI/CD practices. Also, GitLab, Google Cloud and crowdsourcing environments were utilized in the work.
Front- and back-end software development in J2EE-environment using JAVA, JSP and jQuery.