Cracow, Małopolskie, Poland
Passionate about the latest innovating technologies. Advancing my career in the field of Machine Learning.
• Business Application Development: Developed and maintained a Java-based business application responsible for automated log file scanning, parsing, and data transmission to machine learning models. Integrated the machine learning output back into log format for downstream applications and databases. • Cloud Deployment and Management: Hosted and deployed the application on a private cloud, leveraging Openshift technologies. Ensured high availability, fault tolerance, and efficient resource utilization. • Rendezvous Architecture: Collaborated on a challenger-champion model architecture, known as "Rendezvous," to enable seamless communication between the business application and machine learning models in both development and production environments. • Infrastructure Maintenance and Monitoring: Proactively monitored and debugged the Rendezvous architecture using Elasticsearch for log analysis and Grafana for tracking memory utilization. Resolved bottlenecks, ensured optimal operation, and spearheaded the addition of new features to enhance the business application's capabilities.
- Taking part in talks with international teams - Working with other leaders in the company on long-term strategic goals - Taking an active part in internal initiatives - Utilizing recent research and modern practices - Working hands-on on projects in an Agile environment Co-Developed and deployed ML project using NLP techniques and for predicting development groups responsible for fixing problems based on ticket descriptions raised by testers. Model is a mandatory tool used by everyone in Nokia. The model has decreased group misassignments, ticket ping-ponging between groups and time to close reported tickets. • Set up development and production environments using docker containers. • Researched and tested with colleagues multiple models ranging from shallow-methods to deep learning models. • Researched and evaluated various combinations of features and word representations such as sparse vectors and word embeddings. • Continuous extensive corpus cleaning, replacing abbreviations, adding additional context and meaning, etc. • Current working model is a Wide-Deep-CNN and provides the top 5 most probable groups related to the problem. Model is deployed on Nokia’s cloud environment and integrated with Nokia official ticket creation system.
Developed ML project for predicting regression test results by analyzing base station KPI. Effectively replaced static methods used for analyzing results and provides more accurate analysis, decreased analysis time for tester and provided failing root-cause explanations why a test failed. • Responsible for leading a machine learning project - creating, maintaining, optimizing and integrating ML model into automated regression testing pipeline. • Creating a framework in python for gathering, parsing and preparing archived test logs • Selecting, training, and optimizing the best performing machine learning model using open source libraries such as scikit learn. • Setup Spark on multiple servers to form a cluster in order to optimize training speed. • Integrating the optimal model into the automated regression testing pipeline. The model analyzed future executed tests and generated an excel report file showing which samples had passed/failed and which features potentially caused the sample to fail.
• Execute and analyze black box tests on new software features • Execute and analyze automated performance-based regression tests on newly released software builds Create programs/scripts in python for various tasks, such as: o Parsing tests results from database website and comparing current results to previous test results o Searching log files in archived location and zipping all files/folders containing a specific name o Parsing excel file with a list of KPI, checking the list difference with the current database and updating the database with any newly added KPI (with the help of Selenium) o A TV media player program that runs on a Raspberry Pi, controlled from a GUI Web App o Parsing 100GB of stats log files in text format and extracting information, such as how many times a specific error code appeared throughout all the logs