Topic: Framework of the data-driven comprehensive research design applying Machine Learning lifecycle
Comprehensive data-driven research design with ML Lifecycle Framework provides students with the fundamental stages, techniques and tools that enable ML-users to a) prepare, b) test and d) deploy the best practices for managing the complete Machine Learning-based research with scalable, justifiable and repeatable results.
During the lecture you will learn about the following:
- Addressing stages related to monitoring, alerting, and Continuous Integration/Continuous Deployment (CI/CD) and collaboration, issues mitigation.
- Stages for appropriately secured datasets a)design, b)storage, and c)transformations to enable data for specific ML-based research.
- Preparing universal and “ideal” datasets for hypothesis testing and ML-application
- Carrying out experiments with Exploratory Data Analysis (EDA)
- ML Models Management, including AutoML and ML Lifecycle
- Explanations of various Deployment Modalities
- Variations to address Production Issues
- Application of certain heuristics and ideations in vague situations
- Processing up federation, support, versioning and collaboration of the Research and ML Models.
- ML Framework walks students through the end-to-end process of designing a reproducible `Machine learning` Models using customized and/or automated ML (a.k.a.AutoML) solutions based on professionally tailored libraries
All stages of the Framework are accompanied with Code samples and theoretical concepts that were applied in the Research analysis.
Bio
Dmitry Kupov is an Associate, Sr Advanced Analytics Team Lead in Burns & McDonnell who is responsible for designing and supporting Corporate MLOPs/LMsOPs system design. The projects that he is working on in Burns and McDonnell include data and ML models that serve industries like:
– Airspace and Odysseus launching system,
– Nuclear small modular reactors (SMRs), high-temperature gas-cooled reactors (HTGRs), and Renewable energy generation when all of them meet clean energy and decarbonization goals,
– optimization services for the Military Master Planning,
– Oil & Gas manufacturing design and innovative solutions, etc.
Dmitry has his major in applied mathematics and got his Diploma in data science & machine learning with a wide record of executing and managing projects in Machine Learning, Data Engineering, Data Architecture related to predictive analytics and recommendation systems design, MLOPs/LMOPs corporate fine-tuned implementation, and automation. He has more than 15 years of experience in deploying data into comprehensive, justified, durable pipelines where it is turned into corporate knowledge and conventional wisdom across multi-industry companies.
Dmitry devoted his most recent 6 years working with the multi-industry US-based companies that deploy ML solutions into their advanced analytics. He led cross-functional teams in various projects, defining a sustainable and resilient baseline for business units’ development and road maps to achieve corporate deliverables.
Dmitry and his wife Elena with two kids moved to Lexington, Massachusetts four years ago. He enjoys spending time engaging the local community of the towns of Lexington, Sudbury, and Acton in teaching kids extra math curriculum and preparing them for the Math & Computer Science Olympiad competitions.
The team under his supervision took 1st place in the US Continental Math League and won several gold medals in the American Math Olympiad.
