Mark Rudak Email & Phone Number

Lithuania

◾ Ownership of implementing robust system to ensure the reproducibility and testability of machine learning models. This has enabled an organization to make informed decisions quickly and efficiently, reducing decision-making time by 76%.◾ Implemented a production performance degradation service, allowing for monitoring and identification of performance.

Lithuania

• Model predictions injection into the existing or yet non - existing decision making process. This has transformed the organization’s application processing speed from 3 applications per hour to 1 application per second.
• Delivered points of expertise regarding data quality control, estimating direct impact on ROI and turnover. This has resulted in a 50% reduction in operational costs during SME assessment and saved 12,500 hours.
• Developed pipelines for data cleansing, transformation, and generating embeddings, utilizing Feature Stores and schedulers to reduce feature latency by 80%. Lead the whole compliance reporting unit with the team of 5.

Beijing City, China

Being a part of a multinational research team working along on various Machine Learning projects with usage of up-to-date Modelling Technologies and data processing techniques, list of projects that I assisted in:- Utilize AI driven crowd sourcing to enhance the startup success predictions, decreasing FPR by 21%, resulting 1.2x ROI.𝗦𝘁𝗮𝗰𝗸: Python, DS Libs.

Belarus

• Tasks performed:
• Contributed into existing test frameworks (PACT testing, functional API testing,end 2 end tunneling.)
• Created docker configuration for multiple services.
• Reported current status to the customer on regular basis (daily standups).
• Participated in pre-plannings/plannings/retros: analyzed user stories, created tasks for automation, estimated testing time.
• Analyzed and documented defects using TFS.Stack: Java 8, Spring Framework, Docker, Jenkins, Gradle, AWS (EC2, Aurora DB, Elastic Search, SNS)

Belarus

• Optimized database performance and enhance data management processes by means of querying via stored procedures, functions powered by CTEs, VIEWS, the window functions. Redundant data retrieval has been optimized by 80%.
• Streamlined and optimized the data processing workflow, ensuring efficient and accurate data preparation for the ML Feature store, accelerated ML model iterations by 3x.

Belarus

Model-driven data warehousing with data transformations and built-in reporting.

Master Of Data Science And Software Engineering, School Of Software

Bachelor Of Economic Informatics, Economic Informatics